Compare commits

...
19 Commits
Author SHA1 Message Date
Eric Coissac 85296a2ef8 Implement directory locking with shared and exclusive modes
Release / create-release (push) Successful in 2m41s
Release / build-linux-x86_64 (push) Successful in 9m17s
Release / build-macos-arm64 (push) Successful in 2m2s
ci.yml / build (pull_request) Successful in 3m48s
Introduces a robust, OS-level directory locking mechanism using LockSlot for managing shared and exclusive access, supporting lock transitions and non-blocking probes. Also sets up the index structure with new builder and index modules.
2026-10-16 16:02:00 +02:00
Eric Coissac 9ac3a37529 Implement explicit directory locking for index operations
Introduces explicit directory locking mechanisms (`DirLock`) to ensure exclusive access during index operations across command stages. This includes modifying index opening logic to acquire locks before proceeding, handling lock acquisition failures by exiting, and implementing safe conditional building of sibling annexes using locking to prevent concurrent corruption.
2026-10-16 16:00:49 +02:00
Eric Coissac 7c7d7615cb Implement sibling annex building and extension logic
This commit introduces the `SiblingBuilder` and `SiblingExt` traits to manage the creation, invalidation, and atomic building of sibling annexes. It also adds new functionality for sampling, statistics, and improved documentation for sibling-related operations.
2026-10-16 15:59:40 +02:00
Eric Coissac 8296e3544c Simplify Hybrid layer implementation and update dependencies
Release / create-release (push) Successful in 2m31s
Release / build-linux-x86_64 (push) Successful in 9m24s
Release / build-macos-arm64 (push) Successful in 2m5s
ci.yml / build (pull_request) Successful in 3m51s
Dropped fields related to Hybrid layers for simplification and removed the unused `ahash` dependency. Updated dependency versions for `obidebruinj`, `obikmer`, and related crates, and cleaned up temporary artifacts.
2026-10-15 12:23:15 +02:00
Eric Coissac 12afbe819c Refactor MPHF layer logic and evidence handling
This commit implements core lookup functionality for MPHF layers, including `hash` and `find_strict` methods. It introduces support for different evidence types (Exact, Hybrid, Approximate) within layers, clarifying lookup behavior and improving partition calculation logic in the cache. Additionally, dead code is removed and dependencies are updated.
2026-10-15 12:22:45 +02:00
Eric Coissac f9edefd135 Refactor query system for performance and memory reduction
This change introduces dynamic chunk sizing based on input size estimates, ensuring memory constraints are respected while allowing for smaller chunks that yield significant speed improvements. Additionally, index structures were optimized, resulting in drastic memory footprint reduction and faster query execution times.
2026-10-15 12:21:38 +02:00
Eric Coissac 025c40c0db Refactor chunking logic to enforce 128 k-mer limit
This change reduces the maximum number of k-mers per chunk from 256 to 128. This modification requires updating the data model to handle sequences exceeding this limit by splitting them into overlapping chunks, ensuring no k-mers are lost. Extensive testing has been added to validate the new chunking logic and k-mer constraints.
2026-10-15 12:20:36 +02:00
Eric Coissac 8468319a63 Refactor unitig detection and graph traversal logic
Introduce new hashing mechanisms and support for directional unitig checks. Fix fragmentation issues and improve performance metrics through optimized graph construction and new benchmarking utilities.
2026-10-15 12:19:24 +02:00
Eric Coissac d32d86193f Refactor index metadata loading to use buffered I/O
This change replaces unbuffered file reads with buffered I/O operations to improve performance during index metadata loading. This includes implementing new methods for reading index metadata files and wrapping file handles in `BufReader` to optimize subsequent reads.
2026-10-15 12:17:51 +02:00
Eric Coissac 093d81c359 Introduce memory mapping for index files and sparse matrices
This change implements memory-mapped views for index readers and sparse matrices, enabling shared page-cache access. It also includes performance analysis confirming that the default chunk size cap negatively impacts parallelism for small inputs.
2026-10-15 12:17:09 +02:00
Eric Coissac 09c80468b2 Refactor k-mer storage to use flat arrays
This change replaces the HashMap-based k-mer storage mechanism with a compact, flat array structure for `PartitionKmers`. This significantly reduces memory usage per partition and improves query execution time.

It also updates the batch processing logic to handle the new data structure, introduces detailed memory footprint calculations, and updates relevant query layer APIs. Test coverage has been expanded to validate correct partitioning and k-mer ordering.
2026-10-15 12:16:21 +02:00
Eric Coissac fe38424682 Implement batch lookup and chunked data management
Introduces support for batched MPHF lookups, chunked data handling via `SlottedKmers`, and a comprehensive benchmarking suite to measure performance across the indexing and querying pipeline.

This includes new layer functionality for batch operations, chunking logic in the query layer, and scripts to measure wall time and resource usage for various commands. Documentation related to specific implementation details has also been cleaned up.
2026-10-15 12:14:51 +02:00
Eric Coissac 8b87078478 Implement MPHF migration tool and sparse matrix permutation logic
Refactors sparse matrix handling to implement row permutation logic and introduces a standalone tool for migrating indexes from MPHF 1.x to 2.x formats.
2026-10-15 12:12:51 +02:00
Eric Coissac d686d39506 Refactor indexing logic and update core dependencies
This commit introduces significant updates to the indexing system, including the introduction of zero-copy types, modifications to the `Mphf` type definition, and new hybrid search strategies. Core algorithms are updated to handle new hashing types and improved error handling during pointer hash construction. Dependency versions are also updated across the board, including new security features and system-level crates.
2026-10-15 12:07:08 +02:00
Eric Coissac 3bd7fd4f41 Implement memory mapping for file construction
Refactors the building process to utilize memory mapping (`memmap2`) for file creation and introduces the necessary infrastructure for publishing built files atomically. This includes expanding the public API surface and cleaning up deprecated architectural components.
2026-10-15 12:02:33 +02:00
Eric Coissac c226744405 Refactor sampling algorithms and coverage metrics
Optimize family scanning and sampling plan calculation logic. Introduce new per-genome richness metrics and redefine the sampling fraction $r$ for improved analysis.
2026-10-15 12:01:15 +02:00
Eric Coissac 0e40e83fd1 Refactor probability formula notation
Update the mathematical formula for $P(X \geq r)$ to explicitly use summation notation and clearly format the probability expression.
2026-10-15 12:00:51 +02:00
Eric Coissac cb5fc202b6 Refactor sampling logic and introduce diagnostic outputs
This commit introduces new features for sampling, including detailed `SamplePlan` generation, Shannon entropy summaries, and site statistics reporting. It formalizes conditional distance matrix computation and clarifies the logic for family weighting and site selection.
2026-10-15 12:00:44 +02:00
coissac 2232628e58 Merge pull request 'Push kkkmrtvotrsu' (#77) from push-kkkmrtvotrsu into main
Reviewed-on: #77
2026-10-13 07:54:48 +00:00
106 changed files with 7517 additions and 1616 deletions

No files matched your search

+24 -1
View File
@@ -62,7 +62,8 @@ jobs:
~/.cargo/registry
~/.cargo/git
src/target
key: linux-musl-cargo-${{ hashFiles('src/Cargo.lock') }}
tools/migrate-mphf/target
key: linux-musl-cargo-${{ hashFiles('src/Cargo.lock', 'tools/migrate-mphf/Cargo.lock') }}
restore-keys: linux-musl-cargo-
- name: Build static binary
@@ -83,6 +84,28 @@ jobs:
-H "Authorization: token $GITEA_TOKEN" \
-F "attachment=@/tmp/dist/obikmer-linux-x86_64"
# One-shot converter for indexes built with ptr_hash 1.x. A standalone
# crate (own Cargo.lock/target, outside src/'s workspace), built after
# obikmer's own asset is already uploaded: it never blocks that one.
- name: Build migrate-mphf (static)
working-directory: tools/migrate-mphf
env:
PKG_CONFIG_ALLOW_CROSS: "1"
run: cargo zigbuild --release --target x86_64-unknown-linux-musl
- name: Upload migrate-mphf artifact
working-directory: tools/migrate-mphf
env:
GITEA_TOKEN: ${{ secrets.GITEATOKEN }}
RELEASE_ID: ${{ needs.create-release.outputs.release_id }}
run: |
cp target/x86_64-unknown-linux-musl/release/migrate-mphf /tmp/dist/migrate-mphf-linux-x86_64
strip /tmp/dist/migrate-mphf-linux-x86_64
curl -s -X POST \
"${{ github.server_url }}/api/v1/repos/${{ github.repository }}/releases/$RELEASE_ID/assets" \
-H "Authorization: token $GITEA_TOKEN" \
-F "attachment=@/tmp/dist/migrate-mphf-linux-x86_64"
build-macos-arm64:
needs: create-release
runs-on: ubuntu-latest
+1
View File
@@ -9,6 +9,7 @@ CLAUDE.md
memory/
sandbox/
src/target
tools/migrate-mphf/target
data-stress
*.fasta
*.fasta.gz
@@ -0,0 +1,61 @@
# Hot resume and memoization of long computations (discussion, 2026-10-14)
## Goal (user)
1. **Hot resume**: a run killed midway (crash, `Ctrl-C`, scheduler walltime) restarts from the last finished unit of work, not from zero.
2. **No recomputation**: a long function called again with the same parameters (same command rerun, or another command needing the same intermediate) reuses its stored result.
Not a goal: bit-identical resumed samples. Sampling uses an OS-seeded RNG on purpose, repeated runs must stay independent draws to measure variance; a resumed run being a *different* valid draw is not a defect.
## Why it matters: one real run
`phylo --entropy 1 --pseudo-alignment --n-sites 5000000 --free-loss --no-ambiguity --quartet --pair-balance --family-overlap --sibling-stats` on the 180-genome Primulaceae index, from the stderr of the OAR job (2026-10-13), 256 partitions:
| Stage | Wall |
|---|---|
| `sibling_stats` (full scan, `Selection::All`) | 8 235 s |
| `family_overlap` (sweep 1: overlap annex) | 5 582 s |
| `sampled_scan`: sweep 1 again (entropy + quartet annexes) | ≈ 5 347 s |
| `sampled_scan`: sweep 2 (pair-weight annex) | ≈ 5 296 s |
| `sampled_scan`: the draw itself | ≈ 9 573 s |
| `distance` (Jaccard, not asked for; older binary) | 53 s |
| Total | 34 085 s (9 h 28) |
The three annex sweeps cost about the same although their per-family work differs, so the cost is the layer traversal. Part of it was addressed by restricting the matrix pass to selected families (`family_scan.rs`, same session); the sweeps were not fused. Nothing in this run could be resumed or reused: every stage restarts from its beginning.
## What exists today (verified in the code, 2026-10-14)
**Persisted per-layer annexes.** Entropy (`entropy*.pent`), quartet (`quartet*.pbiv`), pair weight, sibling annex: one file per layer, built by `ensure_layer_*` and skipped when `path.exists()` (`family_annexes.rs`). A rerun with the same conventions skips every sweep. This is already memoization, keyed by *file name only* (convention in the name, no fingerprint of the inputs; a rebuilt sibling annex deletes everything derived from it, see `pair_coverage_sampling.md`).
**Not persisted.** The family-overlap accumulator (`FamilyOverlapAccumulator`) is written once at the end of sweep 1 (`sibling_builder.rs`). The draw (`sample_index`) and the `sibling_stats` scan keep their state in memory only.
**Defect found, then fixed (2026-10-14): a killed annex build left a file that looked finished.** `create_mmap_file` (`obicompactvec/src/mmap_file.rs`) created the file at its *final path* with magic, header and the full length (`set_len`) before any data was written. A kill during a layer left a valid-looking file with unwritten data; the next run saw `path.exists()` and treated the annex as built, silently.
Reproduced on the 1x testbed (20 genomes, 512 layers): derived annexes deleted, `phylo --family-overlap --quartet --pair-balance --entropy 1 …` killed by `timeout 6`, then rerun. Old binary: 141 finals after the kill, 2 differing from a reference build; the rerun exited 0 and kept both. Fixed binary: kill at 6 s left 499 finals, all identical to the reference, plus 2 `.building` files; the rerun completed 1 537 files identical to the reference, no `.building` left. (Kill at 3 s: nothing published, nothing broken.)
**Fix.** `obicompactvec::building_path` / `publish_built`: a file is built under `<name>.building` and moved to its final name (same-directory `rename`, atomic) by the builder's `close()`, after the flush. Applied to the builders that go through `create_mmap_file` or write their own mmap on the annex path: `PersistentBitVecBuilder` (all constructors; quartet annex), `PersistentFixedIntVecBuilder`, `PersistentRankSelectBitVecBuilder`, `FamilyScalarAnnexBuilder` (entropy, pair weight), `SiblingAnnexBuilder` (`siblings.psib`) and `FamilyOverlapAccumulator::write` (overlap matrix). A leftover `.building` is truncated by the next build of that file. A build that is dropped without `close()` no longer produces the final file (no caller did; workspace tests pass). Also reordered in `--sibling-annex`: derived annexes are now removed *before* the new sibling annex is built and published, so a kill between the two leaves the old annex with no derived files, never a new annex next to files computed against the old numbering.
**Not covered.** The builders outside that path still write at their final name: `PersistentCompactIntVecBuilder`, the int/sparse matrix builders, `EliasFanoBuilder`, `TempBitVecBuilder`/`TempCompactIntVecBuilder` (temporary by nature) and the index-construction files that use them. The same reasoning applies to them as soon as a stage built on them has to be resumed; not audited. The in-memory overlap accumulator is still lost on a kill (a sweep restarts), now without leaving a corrupt file.
**`obiksession` crate** (`src/obiksession/`): `Session` (named directory, opaque parameter blob checked on open, whole-lifetime `DirLock`, artifacts checksummed with xxh3 and mmap-restored) and `Memoize` (`get_or_compute[_with_key]`, `invalidate`, `cached_keys`, `MemoKey` from `Debug` + xxh3, the `session_memoize!` macro). Compiles, 15 tests pass, **no caller**: `obikmer` lists the dependency without using it. `phylo --session` exits with an error (`cmd/phylo/mod.rs:271`) since `session_cache.rs` (the consumer tied to `PairwiseTally`/`PartitionDispersion`/`SnpAlignment`) was removed. `sample_index(skip_layers)` remains as a seam, never called with a non-zero value.
**Previous design and why it missed the goal** (`theory/evolutionary_distances.md`, section "Explicit `--session`"): one session per *exact* parameter tuple (changing `--n-sites` made another session), artifacts specific to the sampling result, resume only inside the draw (checkpoint every 8 layers or 30 s). Reuse across commands existed only for the sample. An exhaustive-sample cache from which smaller samples would be drawn was proposed and never done. User judgement: this version did not answer the objective.
## Proposed direction (not implemented; to be validated)
Start from the goal, not from the existing crate.
1. **Unit of work = one `(partition, layer)` step that already produces files.** The annex builds, the draw per layer and the scans all iterate that way (`for part … for layer …`). Resume means "skip the units whose output is complete", exactly what `exists()` approximates today.
2. **Complete-or-absent outputs.** Done for the annex path (see the fix above); the remaining builders are to be covered as their stages become resumable. Prerequisite for everything below. The mechanism lives in `obicompactvec` (shared lower-level API), not in each consumer.
3. **Key = function identity + parameters + fingerprint of the inputs.** A result is reusable iff the function, every parameter that influences it and the identity of its inputs (index content, upstream artifacts) match. Today the key is a file name plus an invalidation convention. The fingerprint requirement is the cost: what identifies an index version cheaply is to be decided (meta file, rebuild counter, hash of annex headers).
4. **The cache holds results of long functions, not domain objects.** The crate must not know `PairwiseTally`. Callers serialize their own values (`rkyv` was already chosen for that). Whether `Memoize`'s API fits is unproven: it has no caller, no documentation and was written against the earlier design.
5. **Granularity matters more than the API.** A cached 9 h result helps only the exact rerun; a per-layer cache also helps a changed parameter that leaves earlier layers untouched. The overlap accumulator (end-of-sweep write) would need a per-layer partial (a sum of per-layer matrices, mergeable) to be resumable.
6. **Fuse before caching.** Three sweeps over the same layers where two suffice (overlap first, then entropy + quartet + pair weight together, since the pair weight needs only the finished overlap matrix) is a cost no cache removes on a first run; to do independently.
## Open questions
- Fingerprint of an index version: which cheap, stable identity?
- Overlap accumulator resume: per-layer partial matrices on disk (size n² × layers) or a checkpoint every N layers?
- `Memoize`/`Session`: keep (adapt to the points above), rewrite, or delete the crate until a caller exists (the "existing code is a hypothesis" rule applies; today it has no user).
- Scope of the next step: atomic writes are done for the annex path; extend to the other builders, or go to the cache.
- Draw resume: a layer's draw is independent of the others except through the quotas, which depend on every layer's mass; quotas can be recomputed from annexes on resume (cheap), so only the per-layer survivors need storing.
+43 -1
View File
@@ -24,7 +24,7 @@ for each chunk of sequences (parallel workers via obipipeline, one call to proce
decompose all sequences into superkmers (SuperKmerIter) — construction only,
not the dedup key
deduplicate at k-mer granularity, split by partition in the same pass:
by_partition: Vec<HashMap<CanonicalKmer, Vec<KmerDesc>>> ← KmerDesc = (seq_idx, pos)
by_partition: Vec<PartitionKmers> ← sorted unique k-mers + flat KmerDesc = (seq_idx, pos) slices (2026-10-15, was a HashMap — see "Query memory" below)
allocate SmerIndex (SmerIndex::new): in_index: Vec<bool>, sized total_smers —
NOT multiplied by n_genomes
allocate by_genome: Vec<Vec<(seq_idx, pos, value)>>, one empty Vec per genome —
@@ -404,3 +404,45 @@ Tracked separately in [chunkreader.md](../implementation/chunkreader.md#future-w
- **Thread-budget oversubscription** (phase 4): the single biggest unresolved design question in this whole plan — see phase 4's composition note. Should be settled with real measurements early in phase 4, not assumed from the design alone.
- **`obicompactvec` API surface growth** (phase 4): new public per-column accessors are additive (existing `fill_row`/`row` stay for other callers — `dump`, `select`, distance computations) — no breaking change expected, but worth checking `obicompactvec`'s other callers aren't already relying on `fill_row` being the only/cheapest access path in a way that would make maintaining two access patterns (row-major and column-major) a real maintenance cost rather than a one-off addition.
- **`PersistentBitMatrix::Implicit`'s hardcoded `n_cols: 1` — resolved, not a bug.** `LayerMeta`'s own doc comment (`obicompactvec/src/layer_meta.rs:1-9`) states it is written "alongside `mphf.bin`" and read by `PersistentBitMatrix::open` "to determine `n_rows` for **the implicit (mono-genome presence/absence) case**" — i.e. `Implicit` is a documented single-genome fast path (no presence matrix needed when there is trivially one genome), not a generic "no matrix built yet" fallback. `n_cols: 1` is correct by design for the case it's meant to handle. Phase 4's column loop is safe as planned — this was worth checking once, doesn't need further action.
## Query memory: `PartitionKmers` replaces the per-partition `HashMap` (2026-10-15)
**Problem.** A full-size query (two ~450 MB fastq.gz of a skim against a 354 M-kmer index) reached **74–81 GB RSS** with the default 256 MiB chunk. Cause, read in `batch.rs`/`chunk.rs` and measured with the `by_partition memory retained` debug log: `QueryBatch::by_partition` was a `HashMap<CanonicalKmer, Vec<KmerDesc>>` per partition. On genomic reads almost every k-mer is unique (12.5 M unique for 13.5 M occurrences in a 48 MB chunk, a dedup of 8 %), so each k-mer paid a table slot, growth slack and its own heap `Vec`: **889 MB per 48 MB chunk, ~71 bytes per unique k-mer, 18× the chunk size** — the dominant term of the per-chunk memory, several chunks being in flight at once.
**Change.** `obikquery::PartitionKmers`: occurrences collected as `(kmer, KmerDesc)` per partition, sorted by `(kmer, seq_idx, pos)` and stored as three flat arrays (unique k-mers, u32 offsets, descs). `query_partition_with` takes `&PartitionKmers`; `QueryHit` already carried slices. Occurrence order per k-mer is unchanged (query order); no per-k-mer allocation.
**Measured** (same inputs, release build, identical sorted output digests before and after):
| | `by_partition` (48 MB chunk) | RSS, 2×48 MB chunks | RSS, full run (≈ 1 GB reads) | wall, full run |
|---|---|---|---|---|
| before | 889 MB | 5.6 GB | 73.7–80.7 GB | 135–172 s (one 312 s outlier) |
| after | 259 MB | 4.1 GB | **28.2 GB** | **103 s** |
With `--chunk-size 4` the same small query peaks at 3.6 GB: that floor is the touched pages of the index itself, so after the change the remaining per-chunk anonymous memory is ≈ 6× the chunk size, below the formula's factor 12 (`BASE_OVERHEAD + 8·n_genomes`, still a conservative bound — left unchanged). The sort replaces hashing in `from_records` and was not slower (17.3 s vs 21.2 s on the small test).
**Side finding.** The 256 MiB cap makes small inputs poorly parallel: the two 48 MB files above are two chunks, so two workers; with `--chunk-size 4` the same query takes 3.7 s instead of 17 s. Not changed here.
**Not verified.** Prod-scale (192 workers × 256 MiB chunks in flight): the factor is now ≈ 6 instead of ≈ 21, but how many chunks are in flight at once on that machine was not observed.
### Chunk size sweep on a small input (2026-10-15)
Input: 2 × 48 MB decompressed fastq (200 000 reads each) against a 354 M-kmer single-genome index (4 partitions), 16 cores on AC power, 3 runs per size, sorted outputs identical (same md5) for every size:
| `--chunk-size` (MiB) | chunks | wall (s) | user CPU (s) | RSS (GB) |
|---|---|---|---|---|
| 2 | ~50 | 2.5–2.6 | 25.0–25.3 | 3.2 |
| 4 | ~25 | 2.5–2.7 | 24.3–25.5 | 3.5 |
| 8 | ~12 | 2.5–2.7 | 23.6–26.1 | 4.1 |
| 16 | ~6 | 4.4 | 23.1–23.7 | 4.1 |
| 32 | 4 | 8.3 | 22.9–23.0 | 4.0 |
| 64 and 256 (default) | 2 | 11.6–11.8 | 22.4–22.9 | 4.1 |
Wall time follows the number of chunks (one worker per chunk) until it reaches the core count, then plateaus at ≈ total CPU / cores; total CPU is almost constant (≈ +10 % at 2 MiB, per-chunk fixed cost, here with 1 genome and 4 partitions). With the 256 MiB cap, **any input below ≈ 16 × 256 MiB = 4 GiB decompressed leaves cores idle**. A rule such as `chunk ≈ input_bytes / (workers × 2)` with a floor of 4–8 MiB would use all cores on small inputs — **implemented, see below**. Unmeasured: the per-chunk cost on an index with many genomes and 256 partitions, which could move the floor up.
### Automatic chunk size now follows the input size (2026-10-15)
`cmd/query/mod.rs`: when `--chunk-size` is not given, the memory-derived size (unchanged: `available RAM / (workers × overhead × 2)`, clamped to 4–256 MiB) is further capped by `estimated input bytes / (workers × 2)`, never below 4 MiB. The estimate is `obiread::estimated_decompressed_len`: the file length, ×3 when the magic bytes say gzip (≈ 2.8 measured on FASTQ), ×4 for bzip2/xz/zstd — a deliberately rough figure, detected from the magic bytes, not the extension. For stdin, URLs or unreadable files the size is unknown and the memory-derived size stands. An explicit `--chunk-size` still wins.
Measured (2 × 48 MB reads, 16 cores, AC power, outputs identical): single-genome index, 354 M kmers — default chunk 256 MiB → 4 MiB: **11.4 s → 2.4 s** (−79 %), CPU 22.3 → 26 s (+17 %), peak footprint 1.44 → 0.75 GB. On the 91-genome index the memory-derived size was already 4 MiB (overhead factor 4 + 8 × 91), so nothing changes there (5.1–5.7 s either way). Not measured: a large input (≈ 1 GB gz, whose estimate gives ≈ 95 MiB chunks instead of 256 MiB: more, smaller chunks, expected lower memory).
@@ -0,0 +1,369 @@
# MPHF batch queries — site inventory and expected gains
Investigation of 2026-10-14 (`ptr_hash` 2.1.2, `epserde` 0.13). Question: where would batched MPHF queries (software prefetch of the random accesses) save time? Status of each claim is marked **[read]** (verified in source), **[computed]** (derived from stated sizes) or **[estimate]** (reasoning, not measured). Measurements, once made, are in the last section.
## 1. What `ptr_hash` 2.1.2 actually offers [read]
`Mphf = PtrHash<u64, CubicEps, F = CachelineEfVec, Xxh3Int, V = Vec<u8>, SINGLE_PART = false, REMAP = true>`.
- **`V` is the pilot array** (one byte per bucket), **`F` is only the remap storage**. A layer's `mphf.bin` (~2.4 bits/key **[computed]**: 40 MB for ~130 M keys) is therefore almost entirely a byte array indexed by bucket. The batch APIs prefetch `pilots[bucket]` — the right line. The remap (read only when `slot >= n`, a small fraction of keys) is **not** prefetched.
- `index_batch<K>(xs: [Q; K]) -> [usize; K]` — stable, not feature-gated; hash K keys, prefetch K pilots, query. Remainder (`len % K`) is the caller's job.
- `index_stream<B>(keys: IntoIterator) -> impl Iterator` — ring buffer of B keys; handles head and tail itself; accepts any iterator. **Its `next()` is `unimplemented!`**: only fold-based consumption works (`for_each`, `fold`, `sum`). `collect()`, `zip`, `for x in it` panic. Consume with `for_each` into a pre-sized `Vec`.
- `index_batch_exact` needs the nightly `unstable` feature; `index_batch_exact2` is `#[doc(hidden)]`. Neither is usable.
- Prefetch is `prefetch-index 0.2.1`: `_mm_prefetch(T0)` on x86_64, asm on aarch64, **no-op elsewhere**.
- **`SINGLE_PART`**: neither `index_stream` nor `index_batch` branches on it — both always run the multi-part `bucket()`/`slot()`. Only `index()` has the faster single-part path. So with batched queries the `SINGLE_PART = true` advantage disappears, and `false` (our choice) costs nothing while keeping parallel construction (`true` also panics with sharding).
- Old `ptr_hash` 1.1.0 (used by `tools/migrate-mphf`) has the same family with an extra `MINIMAL` const: `index_stream<B, MINIMAL, Q>`, `index_batch<K, MINIMAL, Q>`. No new dependency needed there.
```rust
// consuming index_stream: fold-based only
let mut out = Vec::with_capacity(kmers.len());
mphf.index_stream::<32, _>(kmers.iter().map(|k| k.raw()))
.for_each(|slot| out.push(slot));
```
## 2. Existing "batch" APIs that are not batched [read]
`MphfLayer::hash_batch` and `MphfOnly::index_batch` loop `.index()` one kmer at a time. Consumers: `obikfilter/partition_iter.rs` (`iter_layer_kmers`), `obikmerge/partition_merge.rs` (pass 2: `hash_batch`, `index_batch`), `obikphylo/siblings/algorithms/family_scan.rs` (two sites).
## 3. Cost chain of a membership query (`MphfLayer::find`) [read]
1. `pilots[bucket]` — the only access `ptr_hash` prefetches.
2. Exact/Hybrid-exact: `evidence.decode(slot)` — 4 B/slot, ~520 MB per 130 M-slot layer **[computed]**, random mmap read.
3. `verify_canonical_kmer` → `block_offsets[...]` (O(1) with the default `--block-size 1`; a forward scan of up to `2^block_bits` records otherwise).
4. `extract_kmer_raw` — random read in the `unitigs.bin` mmap.
Approx/Hybrid-fast: pilots then `fingerprint.get(slot)` — one extra random access.
Each stage is independent across kmers, so a staged batch (slots → prefetch evidence → decode, prefetch offsets/unitigs → verify) can overlap all of them. `ptr_hash` only helps with stage 1; stages 2–4 need prefetch support in `obikindex`/`obiskio` (not existing). Prefetch on an unmapped (cold, not paged-in) mmap page is dropped by the CPU, so none of this helps on a cold index.
Layer sizes: `layer_0` MPHF 2–3 MB (cache-resident), `layer_1` ~40 MB (beyond one CCD's L3 on EPYC-class machines — **hypothesis**, prod CPU not verified, and contended by 192 threads). Gains are expected mainly on `layer_1`. For branch-free independent loops (`hash_batch`) out-of-order execution already overlaps several misses, so the explicit prefetch gains less **[estimate]** than for branchy loops (`find` with early `break`).
## 4. Call-site inventory
(a) already batched; (b) batch buildable; (c) hard/pointless.
| Site | What | Class |
|---|---|---|
| `mphf_layer.rs` `hash_batch`, `MphfOnly::index_batch` | loop of `.index()` | (a) |
| `obikfilter/partition_iter.rs:106` | `hash_batch`, 32768 kmers | (a) |
| `obikmerge/partition_merge.rs:489, 521` | `hash_batch`, `index_batch`, 4096 | (a) |
| `obikphylo/.../family_scan.rs:285, 397` | `hash_batch` | (a) |
| `obikquery/query_layer.rs` stage 1 | `find_in_layer` per kmer, layer by layer, `break` on hit | (b) |
| `partition_merge.rs` `find_batch_grouped` | `find_slot` per kmer over dst layers | (b) |
| `partition_merge.rs` (`build_graph` filter) | `all(find_slot.is_none())`, every layer, per kmer; `build_graph` takes `Fn(CanonicalKmer)->bool` | (b), API change |
| `family_scan.rs` slow path; `annex.rs` | `find_slot` / `find_in_partition` per variant | (b) |
| `mphf_layer.rs` `build`, `build_exact_evidence`, `build_approx_evidence` | `mphf.index` then random writes (evidence, fingerprint, `seen`) | (b), chunked |
| `obikindexer/counter/count.rs:47` | `mphf.index` then `counts.get/set` | (b), hard (needs (kmer, count) pairs) |
| `obikindexer/extensions/private_builder.rs:177` | abundance filter | (b), conditional |
| `private_builder.rs:196`, `verify.rs`, `IndexCache::find_in_*` | per-kmer by nature / sampled | (c) |
| `tools/migrate-mphf/plan.rs:41-42` | `old.index` + `new.index` per kmer, random writes | (b) |
Dead or suspect code found along the way (candidates for removal, not removed):
- `MphfLayer::find_strict` — no caller.
- the single-kmer `hash` chain (`MphfLayer::hash` → `TypedLayer::hash` → `KmerLayer::hash` → `IndexCache::hash`) — no caller outside the chain.
- `IndexCache::find` — no caller.
- `obikmerge/src/merge_layer/` — not declared in `lib.rs` (`mod merge; mod partition_merge;` only); still imports the `LayeredMap` removed in 2026-08. To be confirmed by the compiler.
## 5. Interactions
- **Verification chain**: the right shape is one `find_batch(&[CanonicalKmer], &mut Vec<Option<usize>>)` on `MphfLayer`, staged as in §3, so every consumer benefits (rule: enrich the shared API, no local loops).
- **Memory**: callers already hold bounded batches (4096 / 32768). A buffer-reusing variant avoids one `Vec<usize>` per call. `index_stream` itself only allocates a B-element ring on the stack.
- **Parallelism**: no new parallelism. Batches already come out of `obipipeline` stages; the change lives in the transform body. Construction pass 2 is sequential inside a partition: chunk the streamed iterator (1–4 k kmers), never collect the layer. In `query_partition_with` the unique-kmer `HashMap` is already materialised; iterate it chunk-wise.
## 6. Benchmark
`benchmark/bench_mphf.sh LABEL` times, on the 20-genome benchmark data, `index` (2 genomes), `merge` (20 presence indexes), `query` (presence/count × 2 genomes), `filter`, `convert` (approx/exact round trip) and `phylo` (`--sibling-annex`, `snp-jc`), `REPS` times each, and records order-independent output digests that must not change across labels. See the results section below.
**Limit of this data set**: 20 small genomes → per-partition `layer_1` MPHFs of ~100 KB, fully cache-resident on the dev machine (Apple M3 Max, aarch64). This benchmark can detect regressions and the effect on instruction-level overlap; it **cannot** measure the DRAM-latency gain that motivates prefetch at 40 MB/layer. A faithful test needs a prod-scale layer (see the throwaway benchmark below).
### Throwaway prod-scale benchmark (not written)
Open one real partition's `layer_1` (~130 M keys) and `layer_0` (~8 M) with `MphfOnly`/`MphfLayer`; key sets: present (sample of `iter_kmers`, random order and sequential order) and absent (random, checked absent); measure kmers/s for `index`, `index_stream::<B>` (B ∈ 8…128), `index_batch::<K>` (K ∈ 8…64), and the full `find` vs staged `find_batch`; at 1, 8, 32, 192 threads; pinned to one NUMA node, then spread; warm and cold page cache stated; 5 repetitions, min/median; sizes of `mphf.bin`, `evidence.bin`, `unitigs.bin(.idx)` reported; first check `index_stream` slots == `index` slots. The ~8–12 ns/key figure of the PtrHash paper (quoted in `mphf.md`) is a paper claim, not tied to our sizes or machines.
## 7. Not verified
No measurement at prod scale; prod CPU architecture and cache sizes; evidence mode and `--block-size` used in prod; whether `migrate-mphf` already ran; share of total index/merge/query time spent in these loops; `count.rs` f0 per partition at prod scale; whether `merge_layer/` is dead (only the head of `lib.rs` read).
## 8. First improvement and measurements (2026-10-15)
**Implemented** (query stage 1, the top-ranked site; the same change adds the plumbing the other sites will reuse):
- `MphfLayer` now holds a `MphfOnly` instead of its own `MemCase<Mphf>` (one mmap-open path, one place for the stream call).
- `MphfOnly::for_each_slot(kmers, f)` wraps `index_stream::<32>`; `index_batch`/`hash_batch` now really use it (they previously looped `.index()`). Callback-shaped because `index_stream`'s `next()` is `unimplemented!`.
- `MphfLayer::find_batch` / `TypedLayer::find_slot_batch` / `KmerLayer::find_slot_batch`; the mode-dependent half of `find` is factored into `MphfLayer::confirm`, shared by both.
- `query_partition_with` stage 1 is layer-major over chunks of 4096 unique kmers: each layer receives the still-unresolved kmers as one batch (first hit wins, `n_mphf_calls` unchanged).
- Not done: staged prefetch of evidence/offsets/unitigs (needs a prefetch primitive; `prefetch-index` is already in the tree via `ptr_hash`, but adding it to `obikindex` is a dependency decision), and the other (b) sites.
**Measurements** (`benchmark/bench_mphf.sh`, `benchmark/summarize_bench.py`; Apple M3 Max, aarch64; 20 genomes at 15x). Interleaved A/B, 8 runs each, E. coli reads queried against the 20-genome presence index:
| | wall (s) | user CPU (s) |
|---|---|---|
| before | 5.94–6.02 (median 5.96) | 10.95–11.07 |
| after | 5.63–6.00 (median 5.67) | 10.33–10.66 |
About −5 % wall and −5.5 % CPU, ranges non-overlapping **[measured]**. All output digests (query ×4, merge, filter, convert round trip, phylo) are identical before and after.
**Reading the number.** Small by construction: this data set's `layer_1` MPHFs are ~100 KB per partition, cache-resident, and query time is dominated by reading and k-merising the reads. The gain seen here is instruction-level overlap, not DRAM-latency hiding. The latter, the reason for the whole exercise, remains **unmeasured** and needs a prod-scale layer (§6).
**Noise caveat.** With a load average of 16–22 from unrelated processes, single-run walls of `index`, `merge` and `filter` moved ±10–40 % between identical binaries; only interleaved repeated A/B runs were trustworthy. The `index`/`merge`/`filter`/`convert`/`phylo` rows of the first comparison are noise, not effects (`index` pass 2 was not changed at all; a later interleaved `filter` A/B was inconclusive, 13.0 s vs 13.8 s medians with a 10–20 s spread).
**Stale benchmark indexes.** The indexes already under `benchmark/run/` (`specimen_index_*`, `global_index_*`, built 2026-08-28) carry `ptr_hash` 1.x MPHFs and fail with "Major version mismatch" on current binaries (`tools/migrate-mphf` converts them). `bench_mphf.sh` therefore builds its own copies under `run/bench_mphf/data/` on first use.
## 9. Benchmark procedure
Prerequisites: the simulated reads (`benchmark/run/simulated_data/`) and query reads (`benchmark/run/query_data/`) produced by `gmake simulate simulate_query`; macOS BSD `time -l` and `md5 -q` (the script uses both), Python 3.
1. Build the binary to measure and **copy it** under a name that identifies it (cargo overwrites `target/release/obikmer`): `cp src/target/release/obikmer benchmark/run/bench_mphf/obikmer_<label>`.
2. `BINARY=$PWD/run/bench_mphf/obikmer_<label> benchmark/bench_mphf.sh <label>` (from `benchmark/`; `REPS=3` by default, `CPU_MAX=n` to restrict cores). The first call builds the shared data set in `run/bench_mphf/data/` with that binary (≈ 2 min); keep it for every label — the on-disk format must not differ between the binaries compared, otherwise delete `data/` and rebuild.
3. Output: `run/bench_mphf/<label>/results.csv` (wall and peak RSS per step and repetition, plus one row `step:stage` per Reporter stage) and `checksums.txt`.
4. Compare: `python3 benchmark/summarize_bench.py <ref_label> <new_label>` (medians and ratio) and `diff run/bench_mphf/<ref>/checksums.txt run/bench_mphf/<new>/checksums.txt` — **any difference is a bug**, not a performance matter.
5. A ratio is only a candidate effect. Check the spread by running the **same** binary twice under two labels (the script's own noise floor was ±10–40 % per step on a loaded laptop), or confirm with interleaved A/B runs of the single command (alternate the two binaries, ≥ 8 runs, compare ranges of wall and user CPU from `/usr/bin/time -l`).
6. Rule of thumb for this data set: only `query` (≈ 5 s, stable to ±1 %) resolves effects of a few percent; `index`, `merge`, `filter`, `phylo` need interleaved runs or a quiet machine.
Step → code exercised:
| Step | Code on the MPHF path |
|---|---|
| `index_*` (`:index` stage) | `MphfLayer::build` pass 2 |
| `merge_20_presence` | `find_batch_grouped`, `build_graph` filter, `hash_batch`, `index_batch` |
| `query_*` | `query_partition_with` stage 1 |
| `filter_species` | `iter_layer_kmers` → `hash_batch` |
| `convert_to_*` | `build_exact_evidence`, `build_approx_evidence` |
| `phylo_sibling_annex` | `annex.rs` `find_in_partition` per variant |
| `phylo_snp_distance` | `family_scan.rs` `hash_batch`, `find_slot` |
## 10. Second step — construction pass 2 (2026-10-15)
**Implemented**: `layer/slotted.rs`, `SlottedKmers`, an iterator adapter yielding `(slot, kmer, chunk_id, rank)` for the wrapped `(kmer, chunk_id, rank)` stream, computing slots 4096 kmers at a time through `index_stream::<32>` (bounded memory, never the whole layer). The five loops of `MphfLayer::build` (Exact, Approx, Hybrid), `build_exact_evidence` and `build_approx_evidence` now iterate it instead of calling `mphf.index` per kmer. Output digests unchanged.
**Measured** (interleaved A/B against the first-step binary, 8 runs each, wall / user CPU):
| Command | before | after |
|---|---|---|
| `convert --exact-evidence` (20-genome index) | 0.56–0.68 s / 2.22–2.67 s | 0.59–0.74 s / 2.25–2.52 s |
| `index` E. coli reads | 2.45–2.63 s / 13.1–14.3 s | 2.48–2.75 s / 13.2–14.0 s |
**No measurable effect** on this data set (differences within the run-to-run spread; if anything `convert` is ≈ 4 % slower in wall time). Expected here: the MPHF is cache-resident and `index`/`convert` time is dominated by other stages (scatter, dereplication, counting, unitig decoding, the random evidence/`seen` writes themselves). The construction gain, if any, only exists where the pilot array does not fit in cache, and the random *writes* to `evidence.bin`/`seen`/`fingerprint.bin` (not prefetched by `ptr_hash`) probably dominate there too. Whether to keep this change is therefore a decision to take after the prod-scale benchmark (§6), not on this data.
## 11. Why the two changes gain so little (profiling, 2026-10-15)
Measured with macOS `sample` (1 ms stack sampling) on the 20-genome data, one run each — coarse, but the orders of magnitude are unambiguous.
**Query.** In a 2 s window on the `query` stage (16 threads, ≈ 3 100 active samples against ≈ 55 000 samples of idle waiting):
- `query_partition_with` ≈ 260 samples (8 % of active CPU), of which `find_batch` 217, of which **`verify_canonical_kmer` 185** and everything else in `find_batch` (the `ptr_hash` stream and `confirm`) **32**.
- `verify_canonical_kmer` is the single largest active function of the window (399 top-of-stack samples). The MPHF lookup is ≈ 1 % of active CPU: even a free MPHF could not give more.
- Reason: per-partition pilot arrays are ~100 KB (≈ 25 MB over 256 partitions) and stay cached; the evidence/`unitigs.bin`/offset accesses of the verification (§3, stages 2–4) are the ones that miss. This data set already has the problem on the verification side, not on the MPHF side.
- The threads are mostly idle: the stage is bounded by reading/parsing/k-merising the reads (`rolling_stat`, entropy tracker, string formatting) and by sequential parts, not by lookups.
**What produced the −5 %.** Interleaved A/B, 8 runs each, median wall: original 6.4 s; chunked layer-major restructuring with plain per-kmer `index` (stream disabled) 6.0 s; restructuring + `index_stream` 5.85 s. **Most of the gain comes from the restructuring, not from the prefetch** (the prefetch adds ≈ 2 %, at the limit of the noise). Hypothesis, not isolated: the old loop called `IndexCache::get_layer` — an `Arc<KmerLayer>` clone, i.e. an atomic increment/decrement on a refcount shared by all threads — once per kmer per layer; the batched loop does it once per layer per chunk of 4096.
**Index (pass 2).** In a 1 s window late in the `index` stage, `SlottedKmers` appears with ≈ 7 samples out of 304 per thread (≈ 2 %), against ≈ 200 for the de Bruijn graph work (`obidebruinj`) and ≈ 35 for `ptr_hash::build` (pass 1, MPHF *construction*). Pass 2 is a few percent of `index` and `convert` here: nothing to win even if its lookups were free. `convert` is `build_*_evidence` on 2 MB-scale layers, same conclusion.
**Consequences for priorities.**
1. On this data the next lever is the verification (`verify_canonical_kmer`, evidence decode), not the MPHF — consistent with §3, now observed.
2. Hot `Arc` clones in per-kmer paths (`get_layer`, `IndexCache::find_in_layer`, `find_slot` loops in merge and phylo) are worth auditing independently of MPHF batching.
3. The DRAM-latency gain of prefetch is untested until a prod-scale layer is available (§6).
## 12. Calibration: a single skim at fewer partitions (2026-10-15)
`Vaccinium myrtillus` skim (2 × ~440 MB fastq.gz, `theta 0`), indexed with the current binary, to find the partition count that gives prod-sized MPHFs (`--partitions` is a power of two; MPHF size ≈ 2.4 bits/key **[measured]**):
| `-p` | keys / partition | `mphf.bin` / partition | index wall | peak RSS |
|---|---|---|---|---|
| 256 (default) | ≈ 1.4 M **[computed from p=16]** | ≈ 0.4 MB **[computed]** | – | – |
| 16 | 22.1 M | 6.65 MB | 120 s | 12.3 GB |
| 4 | 86–90 M | 25.9–27.0 MB | 171 s | 19.5 GB |
Total ≈ 354 M distinct kmers for one skim. `-p 4` is already in the 30–45 MB range of the prod `layer_1` MPHFs; `-p 2` would give ≈ 53 MB. Indexes kept in `benchmark/run/bench_mphf/skim_p16` and `skim_p4` (2.3 GB each). Note the index stage itself is dominated by graph and MPHF construction, as in §11.
## 13. Query on a real-scale index (2026-10-15)
**Index.** Working index `phyloskims_sal_vac_sav` (91 genomes of Salix/Vaccinium, 256 partitions × 2 layers, exact evidence, 180 GB, ptr_hash 1.x MPHFs). Converted with `tools/migrate-mphf` **on an APFS clone** (`cp -c`, 0.8 s, no space used until written; the tool works in place and sets the annexes aside, so the original was left untouched — verified by mtime): 512 layers in 10 min (1311 % CPU), ≈ 57 GB of new data. Clone at `/Users/coissac/travail/obiskim/data/phyloalps/bench_mphf/phyloskims_sal_vac_sav_v2`; its sibling annex is in `stale_annexes_pre_ptrhash2/` and would need rebuilding to be used. Per partition: `layer_1` ≈ 31 M keys / 9.3 MB MPHF, `layer_0` ≈ 5.6 M keys / 1.7 MB — larger than the 20-genome set (100 KB) but 4× smaller than the production layers (~130 M keys, ~40 MB).
**Query** (2 × 48 MB of *V. oxycoccos* reads, `--chunk-size 8`, AC power, 3 interleaved runs per binary, identical sorted output for all three):
| binary | wall (s) | user CPU (s) | `query` stage (s) |
|---|---|---|---|
| original | 28.2–30.5 | 70.9–74.4 | 14.6–17.1 |
| + batched query stage 1 | 26.9–28.0 | 48.1–48.6 | 13.6–14.4 |
| + `PartitionKmers` memory fix, build-side `SlottedKmers` | 26.4–26.7 | 41.9–42.5 | 12.9–13.3 |
The `query` stage is ≈ 50 % of wall time; the rest (≈ 13 s) is opening the index (next paragraph). Stage-only gain ≈ −6 % then −11 %; CPU time −33 % then −41 %. **The CPU drop is much larger than the wall drop** and is not explained by MPHF prefetch alone (§11: restructuring, probably the per-kmer `Arc` clones of `get_layer`, not isolated); the original binary was spending CPU in contention that wall time hid.
**Index-open cost (not related to the changes above).** A query of a single 64-base sequence on this index takes **15 s and a 27 GB peak footprint**. `vmmap` during the open: 24.5 GB of `MALLOC_LARGE` in 1 481 regions (≈ 3 per layer, 7.3 GB already swapped out), i.e. heap allocations made by `IndexCache::new` for every layer of every partition whatever the query. Known contributors: `UnitigFileReader` reads each `.idx` into a `Vec` (≈ 30 MB for `layer_1`, ≈ 9 GB over 512 layers); the remaining ~15 GB are in the presence-matrix open path (not yet attributed). Opening only the partitions a query touches is not an option for a read set that spans all of them, but these structures could be mmapped instead of copied.
## 14. Index-open memory: `.idx` offsets and `dict_values` mapped, not copied (2026-10-15)
**Cause** (§13, now attributed): of the 27 GB opened by a one-sequence query on the 91-genome index, **18.2 GB** were `unitigs.bin.idx` read into a `Vec<u32>` **twice** per exact-evidence layer (`MphfLayer::open` builds two `UnitigFileReader`s of the same file — one for the evidence check, one for `iter_kmers`/`find_strict`; both go through `open` → `open_direct_access` → `read_idx` → `fs::read` + copy) and **7.7 GB** were `dict_values.bin` read with `fs::read` in `PersistentSparseBitMatrix::open`. Computed from the real file sizes: 2 × 9.10 + 7.71 = 25.9 GB, against 27.4 GB measured. Everything else the matrix opens (`PersistentRankSelectBitVec`, `PersistentFixedIntVec`, `EliasFano`) was already mmapped.
**Change.**
- `obiskio::unitig_index::reader`: `block_offsets: Vec<u32>` → `BlockOffsets`, a view into the memory-mapped `.idx` (header parsed once, offsets read little-endian on access, no unsafe, no alignment requirement). Truncation is checked at open, so later accesses cannot run past the map. Both readers of a layer now share the same page-cache pages.
- `obicompactvec::bitmatrix::sparse`: `dict_values: Vec<u8>` → `Option<Mmap>` (`None` for an empty file, which cannot be mapped), read through `dict_bytes()`.
**Measured** (same session, same inputs; sorted query outputs identical, md5 `e9e7c35b`):
| | open only (1-sequence query) | 2×48 MB reads, `--chunk-size 8` |
|---|---|---|
| before | 10.6–15.7 s, **27.4 GB** footprint | 27.7 s (51 s cold), 30.9 GB, `query` stage 13.8 s |
| after | **0.26–0.81 s, 0.04 GB** | **10.7 s, 3.6 GB**, `query` stage 9.95 s |
Tests: `cargo test --workspace` green; `scripts/smoke_test_index.sh` PASS; `bench_mphf.sh` (index, merge, query ×4, filter, convert round trip, phylo) completes with all output digests identical to the baseline run, i.e. no failure from files mapped while rewritten (the risk of mapping instead of copying: a reader holding a `.idx` map while the same path is truncated and rewritten would fault). The per-step ratios of that run vs the earlier baseline mix this change with the machine moving from battery to AC power and must not be read as its effect.
## 15. Unbuffered reads: `index.meta` and superkmer files (2026-10-15)
Follow-up of §14 (search for the same family of problems: file content handled inefficiently at open). Static search over `src/` for `fs::read`, `read_to_end`, `load_full`, `from_reader`, bare `File` readers and mmap `to_vec`: every persistent structure of `obicompactvec`, `obikindex` and the `obikphylo` annexes is already mmapped; the two real findings were *unbuffered* readers.
**1. `IndexMeta::read_full` (`obikindex/src/index/meta.rs`)** passed a bare `File` to `serde_json::from_reader`, which issues **one `read` syscall per byte**. `index.meta` of the 91-genome index is 20 KB and is read afresh by every `state()`/`genomes()` call (stateless design, §obikindex audit), so each call cost ≈ 0.35 s here (macOS). Profile of `obikmer dump --head 10` (`sample`): ≈ 20 000 samples in `read`, called from `Bytes<File>::next` ← `serde_json` ← `IndexMeta`. Fix: `fs::read` then `serde_json::from_slice`. **`dump --head 10`: 9.0–10.0 s (91–102 s of system CPU) → 0.16–0.84 s**; the same change applies to `SKFileMeta::read` (`obiskio/src/meta.rs`, per-superkmer-file sidecars).
**2. `SKFileReader::open` (`obiskio/src/reader.rs`)**: for an uncompressed file `niffler::get_reader` returns the sniffed bytes chained to the bare `File`, and `read_superkmer` reads each field with its own small `read_exact`. Fix: wrap the file in a 64 KiB `BufReader` before `niffler`. Effect on `index` (E. coli, 4 interleaved runs each): `count_kmer` 0.51–0.56 s → 0.42–0.50 s (≈ −12 %), total wall unchanged (≈ 1.95 s). Small, low risk, kept.
Tests: `cargo test --workspace` 549 passed / 0 failed; `scripts/smoke_test_index.sh` PASS; query output on the 91-genome index identical (md5 `e9e7c35b`).
**Probes of other commands on the 91-genome index (180 GB, 256 partitions)**: `phylo --distance jaccard` peaks at 155–190 MB of footprint while running and finishes in < 30 s; `utils --stats` 130–150 MB. `filter` (and by the same check `select`, `merge`) refused the index — "source index is not fully built" — with the unmodified baseline binary too: this index's `index.meta` predates the `state` field, so they could not be probed.
**Observations not investigated**: `index` on E. coli shows ≈ 9 s of *system* CPU in 2 s of wall time (both before and after these changes), consistent with the `swtch_pri` (yield) spinning seen in the `index` profile of §11 rather than with I/O; and wall times of the same query on the same binary varied 4.8 s ↔ 10.7 s ↔ 43 s depending on the state of the page cache, so only same-session comparisons are meaningful.
## 16. Profile of `index` on E. coli reads (2026-10-15)
`sample` (1 ms stacks) of `obikmer index` on the E. coli K-12 simulated reads (2 × 15x, 16 threads, one run; the `index` stage lasted 2.4 s under the sampler vs ≈ 0.9 s normally, so proportions are approximate). Stage walls: scatter 0.40 s, dereplicate 0.14, count_kmer 0.60, index 2.36.
**Threads are mostly idle.** Of ≈ 30 800 thread-samples: 24 900 blocked or yielding, 5 900 active. The ≈ 9 s of *system* CPU per run (§15) is `swtch_pri` (`sched_yield`) — 4 550 samples: ≈ 70 % rayon workers spinning for work (`WorkerThread` wait loop), ≈ 28 % `crossbeam` bounded-channel send/recv spinning, the rest in `obidebruinj` thread-locals. It is waiting, not I/O: no `read` samples remain. It costs no wall time on an otherwise idle machine, but burns CPU (visible in `user + sys` accounting and on a shared 192-core machine).
**Where the active CPU goes** (share of the 5 900 active samples):
| | share |
|---|---|
| xxh3 hashing of `CanonicalKmer` for the de Bruijn graph's `HashMap` (callers: `count_neighbors` 791, `hash_one` 358, `walk` 285, `unitig_iter` 252, `push` 164, `unitig_nucleotides` 75, rehash 68; plus `xxh3_stateful_update` 300) | ≈ 34 % |
| `debruijn::count_neighbors` itself (inclusive of its hashing ≈ 2 300 samples: `compute_degrees_and_mark_starts`) | 12.5 % own |
| `ptr_hash::build` (MPHF construction) | 4.8 % |
| crossbeam channel send | 4.8 % |
| `open` / `close` / `write` (many small files over 256 partitions) | ≈ 9 % |
| debruijn `unitig_iter` / `walk` own time | 8.3 % |
So at this scale the cost of `index` is the **de Bruijn graph's hash table** (several lookups of neighbouring k-mers per k-mer, each hashed with xxh3 and probably missing the cache), not the MPHF passes this document started from (`SlottedKmers` ≈ 2 %, §11). Not measured: what fraction of those lookups miss the cache, and the same profile on a prod-size partition.
## 17. Profile of `index` on a skim: *Vaccinium vitis-idaea*, 2.1 GB, `--partitions 16` (2026-10-15)
One run, `sample` windows of 5 s every ~17 s over the whole run (the sampler slows the process; walls below include it): scatter 19.7 s, dereplicate 8.0 s, count_kmer 25.2 s, **index 139.1 s**; total 192 s real, 1830 s user, **225 s system** (12 % of user, against ≈ 90 % on E. coli, §16), peak footprint 17.8 GB.
**Index stage** (samples taken fully inside it, 91 000 active vs 187 000 idle/yielding thread-samples):
| share of active CPU | |
|---|---|
| xxh3 hashing of kmers (de Bruijn `HashMap`) | 32.7 % |
| `debruijn::walk` | 12.4 % |
| `debruijn::count_neighbors` | 10.7 % |
| `GraphDeBruijn::push` | 6.9 % |
| `crossbeam` channel | 6.6 % |
| `hashbrown` table operations | 5.0 % |
| `debruijn::unitig_iter` | 4.8 % |
| memcpy/memset, malloc/free, file I/O, SipHash | ≈ 7 % |
| other | 13.7 % |
- **The de Bruijn graph is ≈ 72 % of the active CPU of `index`; the MPHF construction does not appear (< 1 %)**, and `SlottedKmers`/evidence filling is negligible. The conclusion of §16 holds at 100× the E. coli scale.
- **The stage is poorly parallel**: on average ≈ 5 of 16 threads are active (33 % of thread-samples); `swtch_pri` (spin/yield) is 7 % of all thread-samples. With 16 partitions of unequal size the tail partitions run alone. The 12 % system CPU is the same spinning as §16, at a smaller relative weight.
- **count_kmer** (one window): sorting 40 %, file I/O 18 %, malloc/free 9 %, memcpy 6 %. **scatter** (one window, 32 % active): `RollingStat::push` 32 %, `EntropyTracker::push` 23 % + `normalized_entropy` 5 %, super-kmer stream 10 %, gzip decode ≈ 9 %.
Not established: the cache-miss rate of the graph's table (stalls are attributed to whichever instruction waits, so "hashing 33 %" may include time waiting for the hasher's operands or surrounding loads — a counter-based profile is needed to separate compute from memory stalls), and whether a cheaper hash function for `u64` kmers or fewer lookups per kmer would change the picture.
## 18. De Bruijn graph: hasher, walk, and an orientation bug that fragmented unitigs (2026-10-15)
Benchmark harness: `obikindexer/examples/graph_bench.rs` (`graph_bench <unitigs.bin> <k> [repeat]`; env `IDEAL=1` counts maximal unitigs independently of `GraphDeBruijn`, `SYNTH=n` uses synthetic reads, `DUMP`/`IDEALDUMP` print unitigs). Reference input: the 32.6 M kmers of one partition of the *V. vitis-idaea* index (`-p 16`, k = 31). Phase timings, 16 threads: push / `compute_degrees_and_mark_starts` / `for_each_unitig`.
| | push | degrees | unitigs | unitigs produced | nucleotides |
|---|---|---|---|---|---|
| original | 4.1 s | 2.0 s | 3.9 s | 6.56 M (varies ±900 per run) | 229.4 M |
| `u64` hasher (below) | 1.0 s | 0.7 s | 3.1 s | 6.56 M | 229.4 M |
| + orientation fix | 1.0 s | 0.8 s | **1.0 s** | **3.81 M** (stable ±3) | **146.9 M** |
Independent count of maximal unitigs on the same kmers (nodes − mutually-unique links): 3 809 887. After the fix the algorithm gives 3 809 945 (+0.002 %, cycles).
**1. Hasher** (`debruijn/kmer_hash.rs`). The node table used xxh3's streaming `Hasher` (`Xxh3Builder`), for a key that is one `u64`: ≈ 35 % of the active CPU of `index` (§17). Replaced by a splitmix64 finaliser `BuildHasher` (one multiply–shift pair, no state). push 4.1 → 1.0 s, degrees 2.0 → 0.7 s. `xxhash-rust` is no longer used by `obidebruinj` (left in its `Cargo.toml`; removing it is a dependency decision).
**2. Walk**: `WalkState` now carries the node's own `&AtomicU8`, so marking the next node visited does not look the kmer up a second time (−2 % on the unitig phase: not where the time was).
**3. Unitig fragmentation — a bug, not a tuning.** Starts are detected only for the canonical orientation. A chain whose end node is stored reverse-complemented has no recognised start and falls into the sequential "cycles and tails" pass, where `find_chain_start` returned only the kmer, dropping the orientation, and `unitig_nucleotides` always restarted in the canonical orientation: from the wrong end it extends nowhere and emits a **lone-kmer unitig**, repeatedly, along the chain. The result depended on table iteration order (`degrees_linear_chain_extensions` passed with xxh3's order and failed with any other; counts varied run to run) and was never maximal: **72 % more unitigs than maximal** on the reference partition. A second slip in the reverse branch of `WalkState::walk` (the `can_extend_left`/`can_extend_right` test of the next node was swapped) cost another ~15 %. Fixes: `walk` condition corrected; `find_chain_start` returns `(kmer, direct)` with the orientation to walk forward; `unitig_nucleotides` and `UnitigNucIter` take that orientation (and read the first k nucleotides reverse-complemented when it is false); starts are now flagged in both orientations and the claim picks the right one (`start_direction`). Regression tests: `linear_chain_is_one_unitig_whatever_the_orientations` (fails on the original code, passes now) and `unitigs_partition_the_kmers_of_branching_reads` (every kmer in exactly one unitig).
Residual: on noisy synthetic reads (k = 11–31, 200–20 000 bases) the algorithm still gives up to 5 % more unitigs than the independent count, always a few short pieces next to chain ends; not explained, absent from real data (+0.002 %).
**End-to-end** (current binary vs previous one, same inputs):
| | before | after |
|---|---|---|
| *V. vitis-idaea* 2.1 GB, `--partitions 16`: total wall | 163.8 s | **116.5 s** (−29 %) |
| … `index` stage | 114.6 s | **59.2 s** (−48 %) |
| … user CPU | 1 870 s | 1 304 s |
| … `unitigs.bin` / `.idx` | 1.05 / 0.42 GB | **0.66 / 0.24 GB** (−37 % / −43 %); index 3.5 → 2.9 GB |
| E. coli, `index` stage / wall / user CPU | 0.91 s / 1.97 s / 10.2 s | 0.59 s / 1.63 s / 7.3 s |
Same k-mers: per-layer kmer counts identical (521 630 625 over the 16 layers), E. coli `dump` and `query` outputs identical (md5), `scripts/smoke_test_index.sh` PASS, `cargo test --workspace` 551 passed / 0 failed, and `bench_mphf.sh` digests (index, merge, query ×4, filter, convert round trip, phylo) identical to the baseline run. Existing indexes are unaffected (any partition of kmers into unitigs is valid; evidence records are per-index); only newly built layers get the shorter, fewer unitigs.
## 19. Rank field: unitig chunks capped at 128 k-mers (2026-10-15)
`Evidence` packs `(chunk_id, rank)` into a `u32` with a 7-bit rank (`encode` masks `& 0x7F`), but `MAX_KMERS_PER_CHUNK` was 256 (the width of the unitig record's length byte), so a chunk of 129–256 k-mers had its ranks ≥ 128 silently truncated and those k-mers no longer verified. Observed on E. coli reads, same query: **`--partitions 256`** → max chunk 49–50 k-mers, results correct with the old and the new binary; **`--partitions 1`** → max chunk 256, query output wrong with both the old (md5 `bde0f886a0`) and the post-graph-fix (`8f1c94ef32`) binary versus the correct `f99e971087`. The orientation fix of §18 makes unitigs longer, hence makes this more likely at low partition counts.
**Change** (`obikseq::MAX_KMERS_PER_CHUNK`: 256 → 128; chosen so, rather than failing at write time: a longer unitig is simply written as consecutive chunks, nothing is lost). The constant is shared with the super-kmer chunking of the scatter files (`SuperKmer::write_to_binary`): super-kmers above 128 k-mers (rare) are now split into 128-k-mer records — readers are unchanged and old files stay readable. Tests that hard-coded 255/256/257/259/260 were made relative to the constant; new test `chunk_ranks_fit_the_7_bit_evidence_field`.
**After** (E. coli, same query): `--partitions` 256 / 16 / 4 / 1 → max chunk 50 / 70 / 118 / 128 k-mers, query md5 `f99e971087` in all four cases. `cargo test --workspace`: 551 passed, 0 failed.
**Existing indexes.** An index built with 256 partitions has chunks far below the limit (91-genome index, first 4 partitions: max 42–56, none above 128) and is unaffected. The `--partitions 4` skim index used for the benchmarks above (built before the cap) holds 6 chunks over 128 k-mers in its 4 partitions (max 135): a handful of k-mers out of ≈ 354 M verify wrongly. `scripts/check_unitig_chunks.py "<index>/partitions/part_*/index/layer_*/unitigs.bin"` reports chunks over 128 (it reads every unitigs.bin sequentially; k is hard-coded to 31 at its top).
## 20. Summary of what was gained (2026-10-15)
Apple M3 Max (aarch64, 16 cores, 128 GB), on AC power; absolute times are not those of the 192-core production machine, ratios measured in one session are what is reported. "Original" = the binary at the start of this work; "now" = current tree. Every output digest checked (query, dump, merge, filter, convert, phylo) is identical before and after, except where stated.
### Speed
| Operation | Original | Now | Where |
|---|---|---|---|
| Open an index (query of one 64-base sequence), 91 genomes, 180 GB | 10.4–13.3 s | **0.18–0.19 s** | §14, same-session A/B |
| Query, 2 × 48 MB reads on that index (`--chunk-size 8`) | 30.7–32.0 s (61.8 s cold), 70–76 s CPU | **11.6–13.2 s**, 36–39 s CPU (−62 % wall, −50 % CPU) | §14, same-session A/B |
| `dump --head 10` on that index | 9.0–10.0 s (91–102 s system CPU) | **0.16–0.84 s** | §15 |
| `index` of a 2.1 GB skim (*V. vitis-idaea*, `--partitions 16`) | 163.8 s (stage `index` 114.6 s) | **116.5 s** (stage `index` 59.2 s): −29 % wall, −48 % on the stage, −30 % CPU | §18 |
| `index` of E. coli reads | 1.97 s (stage 0.91 s), 10.2 s CPU | 1.63 s (stage 0.59 s), 7.3 s CPU | §18 |
| De Bruijn graph, 32.6 M kmers (push / degrees / unitigs) | 4.1 s / 2.0 s / 3.9 s | 1.0 s / 0.8 s / 1.0 s (×3.5 overall) | §18 |
| `index` stage `count_kmer` (E. coli) | 0.51–0.56 s | 0.42–0.50 s | §15 |
| Query on a small input (2 × 48 MB reads), single-genome index | 11.4 s (2 chunks = 2 busy cores) | **2.4 s** (−79 %) — automatic chunk size now follows the input size (`architecture/query.md`); on the 91-genome index the size was already 4 MiB | §20 follow-up |
### Memory (peak RSS / footprint)
| Operation | Original | Now |
|---|---|---|
| Opening the 91-genome index (any command that builds an `IndexCache`) | **27.4 GB** | **0.04 GB** (§14: `.idx` offsets and `dict_values.bin` mapped instead of copied; they were read twice and once) |
| Query, 2 × 48 MB reads on that index | 32.2–32.5 GB | **3.5–3.6 GB** |
| Full query, ≈ 1 GB gz of reads, 354 M-kmer index, default chunk size | **73.7–80.7 GB** | **28.2 GB** (§12 of `architecture/query.md`: `PartitionKmers` instead of a `HashMap<kmer, Vec>` per partition: 71 → ~20 bytes per unique kmer) |
| `index` of the 2.1 GB skim | 19.7 GB | 18.3 GB |
### Index size (newly built indexes; existing ones are unchanged)
| | Original | Now |
|---|---|---|
| `unitigs.bin` / `unitigs.bin.idx` (2.1 GB skim, 521.6 M kmers, 16 partitions) | 1.05 / 0.42 GB | **0.66 / 0.24 GB** (−37 % / −43 %: 3.81 M unitigs instead of 6.56 M — the orientation fix of §18) |
| Whole index of that skim | 3.5 GB | **2.9 GB** (−17 %) |
| `evidence.bin` | 4 bytes per kmer | unchanged (still ≈ 40 % of the 91-genome index, 2.1 GB of 3.5 GB for the skim): reducing it is a separate design question (§ answer on exact evidence: ≈ −22 to −31 % of it by dropping the rank field, far more with a minimizer-indexed layout; neither implemented) |
### Correctness fixes found on the way
- Unitigs were not maximal (72 % too many) and depended on hash-table iteration order (§18).
- Chunks of 129–256 k-mers had their ranks silently truncated in the 7-bit evidence field and failed verification, visible with few partitions (§19, `scripts/check_unitig_chunks.py` to audit an index).
- `index.meta` and super-kmer files were parsed byte by byte (§15).
### What is **not** established
- The DRAM-latency gain of MPHF prefetch itself, the original motivation (§1–§6): the measured MPHF share of the profiles is ≈ 1 % (§11), and the layers here are 4× to 1 000× smaller than production ones. The −5 % seen on the first query change came mostly from batching per layer (probably fewer shared `Arc` clones, §11), not from prefetch. A prod-scale benchmark (§6) remains to be run on the production machine.
- Speed-ups on the 192-core / 1 TB production machine, on `merge`, `filter`, `select` and `phylo` at the 180-genome scale (the 91-genome index could only be used for `query`, `dump`, `phylo --distance jaccard`, `utils --stats`; `filter`/`select`/`merge` refused its pre-`state` `index.meta` until it was patched in the clone).
- The size of the Primula index composition (§ calculated: ≈ half evidence from 35–40 G kmers × 4 bytes; not measured).
## 21. Dead code removed (2026-10-15)
The items listed as unused in §4 were removed after Serena's `safe_delete_symbol` and the compiler confirmed no reference: `MphfLayer::find_strict`; the single-kmer `hash` chain (`MphfLayer::hash`, `TypedLayer::hash`, `KmerLayer::hash`, `IndexCache::hash`); `IndexCache::find`; the uncompiled `obikmerge/src/merge_layer/` directory (689 lines; `lib.rs` only declares `merge` and `partition_merge`); and `xxhash-rust` from `obidebruinj/Cargo.toml` (it still serves `obikseq`, `obifastwrite` and `obiksession`).
Side effect: `find_strict` was the only reader of the `unitigs` handle that `Hybrid` layers opened, so that field (and the extra `.idx` map and mmap per Hybrid layer) was dropped too. `find` on a Hybrid layer was already a fingerprint check; the exact evidence stays on disk (it gives the slot count and is what `convert` works from) but no API consults it any more. Restoring an exact lookup for Hybrid layers means re-adding `find_strict` and that field.
`ahash` was not referenced by any source file of `obidebruinj` either and was removed from its `Cargo.toml` too.
Checks after the removal: `cargo build --release` without warnings, `cargo test --workspace` 555 passed / 0 failed, `scripts/smoke_test_index.sh` PASS.
Scratch artifacts of this work were deleted: the intermediate binaries, the two skim indexes (`skim_p4`, `skim_p16`) and the shared data set `benchmark/run/bench_mphf/data` (rebuilt automatically, ≈ 2 min, by the next `bench_mphf.sh`). Kept in `benchmark/run/bench_mphf/`: the original binary `obikmer_baseline` (reference for A/B runs) and the per-label results and checksums (18 MB in all). The 91-genome index converted to the current MPHF format (`/Users/coissac/travail/obiskim/data/phyloalps/bench_mphf/phyloskims_sal_vac_sav_v2`, with its `index.meta` patched, original `index.meta.before_state_fix` beside it) was left in place.
+4 -4
View File
@@ -22,10 +22,10 @@ A sequence of binary records. Each record:
- Packed nucleotides: A=00, C=01, G=10, T=11, MSB-first within each byte; last byte zero-padded.
- Byte count for packed sequence: `ceil(seql / 4)`.
Unitigs with more than `MAX_KMERS_PER_CHUNK = 256` k-mers are transparently split into overlapping chunks. Each chunk has at most 256 k-mers (= `seql − k + 1 ≤ 256`); consecutive chunks overlap by k−1 nucleotides so no kmer is lost:
Unitigs with more than `MAX_KMERS_PER_CHUNK = 128` k-mers are transparently split into overlapping chunks. Each chunk has at most 128 k-mers (= `seql − k + 1 ≤ 128`); consecutive chunks overlap by k−1 nucleotides so no kmer is lost:
```
chunk 1: nucleotides [0, MAX_KMERS_PER_CHUNK + k − 2] (256 kmers)
chunk 1: nucleotides [0, MAX_KMERS_PER_CHUNK + k − 2] (128 kmers)
chunk 2: nucleotides [256, end] (remaining kmers)
overlap: k−1 nucleotides shared between the two chunks
```
@@ -95,11 +95,11 @@ Two memory accesses: one 4-byte read from `evidence.bin`, one packed-bit extract
| field | bits | range | capacity check (*B. nana*, 256 partitions) |
|------------|------|---------------|---------------------------------------------|
| `seql − k` | 8 | 0–255 | max `n_kmers` per chunk = 256 = `MAX_KMERS_PER_CHUNK` |
| `seql − k` | 8 | 0–127 used | max `n_kmers` per chunk = 128 = `MAX_KMERS_PER_CHUNK` (the field could carry 256; capped so that every rank fits 7 bits) |
| `rank` | 7 | 0–127 | observed max ~46 kmers/chunk; structural max k−m+1 = 21 |
| `chunk_id` | 25 | 0–33 554 431 | avg U ≈ 275 k chunks/partition |
The rank field is 7 bits (max 127) even though chunks can contain up to 256 k-mers, because rank counts only canonical kmers within the chunk, and the canonical kmer count is at most half the total.
The rank field is 7 bits (max 127) and the rank of a k-mer is its index among the chunk's k-mers, so a chunk must hold at most 128 of them: that is what `MAX_KMERS_PER_CHUNK = 128` guarantees. (Until 2026-10-15 the cap was 256 and the earlier justification here — that rank counts only canonical k-mers, at most half the total — was wrong: `iter_indexed_canonical_kmers` yields one entry per position. Chunks of 129–256 k-mers occur with few partitions — 6 of them in a `--partitions 4` index of a 2 GB skim, none with 256 partitions where the observed maximum is ≈ 50 — and their k-mers beyond rank 127 were silently stored with a truncated rank. `scripts/check_unitig_chunks.py` finds such chunks in an existing index.)
---
+81 -1
View File
@@ -1,4 +1,4 @@
# Pair-coverage-biased family sampling (discussion, 2026-10-07)
&@")--# Pair-coverage-biased family sampling (discussion, 2026-10-07)
Idea: bias the family sample used by `--pseudo-alignment`/`--sankoff`/`snp-*` so that the `family_overlap` matrix of the *sample* is as uniform as possible, compensating the scarcity of shared families between distant species. Complements, does not replace, the existing proportional-by-partition and entropy-biased draws (see user doc, *Sampling design*).
@@ -278,6 +278,86 @@ The weight flattens the representation (slope toward 0) and, above all, covers m
Reading: `--quartet` is the decisive option here — a uniform draw of 100-300 sites contains almost no informative column (0.5-2), the restricted draw contains only informative ones. `--pair-balance` alone barely helps (it favours rare pairs but does not remove the uninformative columns); added to `--quartet` it raises the share of replicates recovering all three species (50→53 % and 33→67 %). The entropy kernel (centre 1 bit, default width) adds nothing at 300 sites and costs at 100. Limits: 11 genomes, three simulated 1x datasets, small budgets, a single criterion (species monophyly, not deep branches), and the differences between the `--quartet` variants (a few runs out of 30) are small relative to the replicate noise — a direction, not a ranking.
## Implemented: no distance matrix unless one is wanted (2026-10-13)
`--distance` had a default (`jaccard`) and `phylo` always computed and wrote a matrix, even for `--sibling-annex`, `--family-overlap` or `--pseudo-alignment` alone. `--distance` is now optional: a matrix is computed only if `--distance` is given, or `--nj`/`--upgma`/`--shared-kmers` needs one (then `jaccard`, `args::distance_to_compute`, unit-tested); `--csv` alone changes nothing. When nothing is wanted the command stops before the distance step. Checked on the 1x testbed: `--family-overlap` and `--pseudo-alignment` runs write no `_dist` file and run no distance stage; `--nj` writes `_dist.phy` and `_nj.nwk`; `--distance mash --csv` writes `_dist.csv`. Cost saved on a large index: not measured.
## Implemented: diagnostic outputs (2026-10-13)
Added so the measurements this file reports no longer need ad hoc scripts, and so a draw can be diagnosed from the files of the run that produced it:
- `--sample-stats`: a `SampleStatsReducer` (`algorithms/sample_stats.rs`), a `LayerReducer` run beside the alignment over the same surviving families; writes `_sample_genomes.csv`, `_sample_carriers.csv`, `_sample_overlap.csv`. A call is a masked state other than `0`; informative = `has_bipartition` (the quartet rule, now shared). Checked on the 1x testbed: per-genome calls and the overlap matrix equal those recomputed from the alignment.
- `--sample-plan`: `sampled_scan` now returns a `SamplePlan` (one `LayerReport` per layer: eligible, mass, quota, resolved, kept); `_sample_plan.csv` has the layer lines only (quotas sum to the target by cumulative rounding; on the testbed 3000 asked, 2695 kept).
- `--shannon-summary`: `--shannon` and the summary share one traversal (`scan_family_entropies`, which replaces the loop that lived in `shannon_entropy_csv`); the summary keeps counts per (entropy kind, genomes present, genomes without ambiguity, 0.05-bit bin) instead of one line per family: 156 lines for the 46 706 families of the testbed (a few hundred thousand lines at most on any index, by construction: the bins are bounded). Checked: its `entropy15` counts sum to the number of variable families.
- Writers are in `cmd/phylo/stats.rs`; the library computes, the command layer formats and writes.
## Combined draw on Primulaceae: `--quartet --pair-balance --entropy 1` (measured 2026-10-13)
Command: `phylo --entropy 1 --pseudo-alignment --n-sites 5000000 --free-loss --no-ambiguity --quartet --pair-balance -o primulaceae_quartet_balance --family-overlap --sibling-stats primulaceae`. One draw, 180 genomes, 5 000 000 columns requested and obtained (no shortfall; no `--sample-plan`, so per-layer quotas are not inspectable). The three options are on together: **effects of entropy, quartet and pair balance are not separable here**; the testbed result that the entropy kernel adds nothing at small budgets was not rechecked.
**Baseline.** The same statistics were recomputed with one script on the alignment fed to IQ-TREE (`primulaceae-gamma_alignment.fasta`, 5 003 414 columns). It reproduces the figures of the earlier sections (sample overlap min 4 / median 1 665 / max 387 057 / CV 1.57; sampling fraction $r$ quantiles 0.0025 / 0.0037 / 0.0055; 5.2 % median calls per genome), hence **is the entropy-drawn sample** (`--entropy 1 --entropy-sd 0.5`, section on entropy bias), not a uniform draw. The "uniform draw" label of the nesting section is inconsistent with this (to settle). So the comparison below is *entropy-only vs entropy + quartet + pair balance*, never against a uniform control. A call is any character other than `?`/`-`. With `--free-loss` absence is written `?` (`masking::masked_state`); `-` only occurs without it. Neither file contains IUPAC characters.
| | Baseline (entropy only) | Combined |
|---|---|---|
| Informative columns (≥ 2 bases, each in ≥ 2 genomes) | 21.4 % | 100 % |
| Carriers per column: median / share ≤ 4 / share ≥ 33 | 4 / 58.1 % / 8.0 % | 8 / 12.4 % / 11.3 % |
| Calls per genome, median; genomes < 1 % / < 2 % | 5.2 %; 8 / 23 | 8.1 %; 4 / 6 |
| Sample pair overlap min / median / max | 4 / 1 665 / 387 057 | 282 / 27 380 / 491 038 |
| Sample overlap CV; max/min | 1.57; 96 764 | 1.10; 1 741 |
| 4-way count (20 000 random quartets, 2 seeds): median | 200-202 | 3 738-3 797 |
| 4-way p1 / p10 | 3 / 66-67 | 93-95 / 1 092-1 104 |
| 4-way p90 / p99 | 3 000-3 644 / 65 700-67 300 | 14 500-15 600 / 82 900-84 600 |
| 4-way CV; share of quartets with 0 | 3.4-3.5; 0.11-0.14 % | 1.83-1.87; 0 % |
(p90/p99 move by 20 % between the two seeds: read them as orders of magnitude only.) The gain is at the bottom of the 4-way distribution (median ×19, p1 ×30); the top is unchanged. 4-way coverage is not stratified by clade depth here (needs the tree). "Informative" is measured over the 180 genomes: a column informative overall is not necessarily informative among the four genomes of a given quartet.
**Pair representation.** Index-wide overlap used for the comparison: baseline, the raw matrix (`primulaceae_phylo_family_overlap.csv`); combined, the `--no-ambiguity` matrix of the same run. Different conventions, so the two slopes are indicative.
| | Baseline | Combined |
|---|---|---|
| Slope of log(sample overlap) on log(index overlap) | 1.12 ($R^2$ 0.997) | 0.54 ($R^2$ 0.91) |
| Pairs with index overlap > 0 and no sample column | 0 | 0 |
| Median $r$ = sample/index, poorest decile of index overlap | 0.0027 | 0.059 |
| Median $r$, richest decile | 0.0053 | 0.0067 |
| Median $r$ by decile 1 → 10 | 0.0027 … 0.0053, rising | 0.059, 0.065, 0.067, 0.067, 0.069, 0.074, 0.023, 0.0082, 0.0077, 0.0067 |
The weight does not merely flatten: the poorest pairs are now about 9× better represented (relative to their pool) than the richest, where the baseline gave them about half. Deciles 1-6 are flat; the drop sits at deciles 7-10, around the gap of the index-overlap distribution already noted in the entropy-bias section (cause of that gap still unchecked). **Caveat on $r$:** the numerator only holds quartet-informative families while the denominator counts every variable family, and the matrices differ in convention, so $r$ is no longer a sampling fraction of one pool; the decile pattern is a measurement of the sample, not of an inclusion probability.
**Per-genome richness.** `--sibling-stats` now gives the variable families (size ≥ 2) per genome: 0.72 M to 93.9 M, median 34.5 M. Spearman between a genome's calls in the sample and its variable-family count: **0.974 in the baseline, 0.435 in the combined draw**: the baseline mirrored genome richness, the combined draw largely decouples from it (on the same genomes the correlation with *all* families is 0.66 and 0.32). Genomes poorest in variable families: *A. saussurei* 0.72 M (0.30 % of its families), *A. triflora* 1.26 M (0.72 %), *P. stricta var. obesior* 2.39 M (1.33 %), then *P. pumila* 5.2 M, *P. auricula* subsp. *serratifolia* 5.9 M (1.92 %), *P. carniolica* 6.8 M (1.77 %), *A. pyrenaica* PHA000543 8.4 M (1.46 %). Their calls in the sample went from 6 986 / 1 936 / 2 403 (saussurei, triflora, stricta) to 15 725 / 22 988 / 25 392, and carniolica from 4 937 to 52 310; *saussurei* stays the poorest genome in both. These genomes are data-poor, not empty: their families are mostly of size 1. This is the per-genome $c_i$ the normalisation of the data-poor section calls for.
Global family-size histogram (`global` row): 34 047 616 087 families; sizes 1 / 2 / 3 / 4 = 33 512 358 231 (98.43 %) / 479 910 992 (1.41 %) / 45 621 663 (0.13 %) / 9 725 201 (0.03 %). Sizes ≥ 2 sum to 535 257 856, matching the 535 257 857 of the entropy table within 1.
**`_dist.phy` (not asked for).** 180 × 180, symmetric, values 0.456-0.99999, median 0.999; Spearman with the Mash matrix 0.9999998 and with the index overlap −0.95, so it carries Mash-like information, consistent with the `jaccard` default (not verified). `args::distance_to_compute` at HEAD computes no matrix for this command line; why the file exists is unknown (binary predating the change, or a leftover; not checked). Not used above.
**Not measured:** any tree-level effect (sCF, species monophyly, support at the interface nodes 209/216 where `sN` was 17 and 7); the separate contribution of entropy, quartet and pair balance (needs ablation runs: without `--entropy`, without `--pair-balance`, `--quartet` alone); variance between draws (one draw per setting); clade-stratified 4-way coverage; the new alignment had not been given to IQ-TREE (done 2026-10-16: see the last measured section).
## Tree-level effect of the combined draw: sCF on three trees (measured 2026-10-16)
Closes the "not measured" item of the previous section (the combined alignment given to IQ-TREE). Three trees, all sCF by `--scfl 100 --cf-verbose`, model `GTR+G4+ASC`:
- **A** = `concord` (12/10): entropy-only draw, 180 genomes, 5 003 414 columns; `-t` + `-m` re-ran a full ML search (59 h).
- **B** = `concord_quartet_balance`: combined draw (`--entropy 1 --quartet --pair-balance`), 180 genomes, 5 000 000 columns. Its log stops during the tree search (14/10), so the command that produced its `cf.*` (16/10) is not in the files; presumably `-te` on the 15/10 treefile as for G (unverified).
- **G** = `scf_good`: same draw as B without the 3 data-poor genomes (*A. pyrenaica* PHA000543, *A. saussurei*, *A. triflora*), 177 genomes. 72 sites became invariant and were removed to satisfy ASC; the search (`concord_quartet_balance_good`, 107 iterations, 6 h 47 on 120 threads) crashed in the sCF step on the FASTA (`Assertion all_prob_const < 1.0`; cause untested, ASC with constant sites is a suspect). sCF were recomputed with `-te <treefile> -s <varsites.phy> -m GTR+G4+ASC --scfl 100`: **no search** (0 iterations, lnL −41 053 676.53 vs −41 053 676.41 from the search), 4 999 729 columns. A used `-t`, B/G a fixed tree: the protocols differ.
| | A | B | G |
|---|---|---|---|
| Branches | 177 | 177 | 174 |
| Mean `sN` | 3 219 | 13 163 | 13 396 |
| Branches with `sN` < 100 / < 1 000 | 13 / 50 | 0 / 5 | 0 / 2 |
| Median sCF | 44.2 | 42.9 | 42.8 |
| Branches with sCF < 40 / ≥ 70 | 74 / 28 | 67 / 26 | 69 / 25 |
| *Primula* and *Androsace* monophyletic | no / no | no / no | yes / yes |
**Sampling effect (A → B, same 180 genomes).** `sN` ×4.5 and the scarcity is gone. Sites do not become more concordant: on the 113 shared bipartitions (of 177) median sCF 48.0 → 47.3 (44 up, 69 down). Robinson-Foulds 128: the draw rearranged the tree deeply. Consistent with the earlier reading that the intra-*Androsace* conflict is not a matter of site count; the number of branches with sCF < 40 % stays around 70.
**Interface nodes.** Of the old nodes 209/211/212/216/217 (`sN` 17 / 226 / 184 / 7 / 7) none exists in G. Node 208 survives with `sN` 190 → 5 218 (sCF 54.5 → 37.5). Six mixed clades remain in G (83 to 72 tips, 50 *Primula* + 22-33 *Androsace*), all with `sN` > 3 000.
**Effect of the 3 genomes (B → G, same draw).** RF 16 (166 of 174 shared), median sCF 43.2 → 43.6, median `sN` unchanged. In B, *A. pyrenaica* and *A. triflora* sit inside *Primula* (a clade of 52 tips = 50 *Primula* + these 2; each has a *Primula* sister, sCF 92 % and 52 %, `sN` 556 and 153); *A. saussurei* stays with *Androsace*. Removing them restores both monophylies and changes nothing else measurably: support for "too little data to place them" (data-poor section), not for contamination, though the latter is not excluded.
**Where A and G disagree (2026-10-16).** On the 177 common tips: 124 shared bipartitions, 50 only in A, 50 only in G. Differing bipartitions are weakly supported in both trees (median sCF 35.7 in A-only, 37.7 in G-only, versus 47.5-48.1 on the shared ones), so the disagreement sits where support was already low. Differences are mostly local: the closest bipartition of the other tree is 1 tip away for 25 of 50 A-only and 21 of 50 G-only splits (≤ 2 tips: 37 and 30); only 8 and 9 are 5+ tips apart. Greedy removal of the tip that lowers RF most: the first (*P. auricula* subsp. *serratifolia*) takes RF from 100 to 86, then about 2 per tip (*A. maxima*, *P. stricta* var. *obesior*, then ordinary tips); the top three are data-poor genomes. A bipartition separates a data-poor genome (the 5 left after the 3 removals) in 1 of 124 shared splits but in 23 of 50 A-only and 10 of 50 G-only splits; removing those 5 genomes gives RF 72, against 96.6 (range 90-100, 40 draws) for 5 random tips. So about 28 of the 100 differences trace to data-poor genomes, and the remaining 72 do not. Backbone: A has a clade of 50 *Primula* + 25 *Androsace* (sCF 76, `sN` 453), G one of 50 + 26 (sCF 73, `sN` 6 677): one tip apart. Genuine conflict: a clade of 72 *Androsace* exists in both trees (A: sCF 34, `sN` 17 127; G: sCF 38, `sN` 49 473) with 16 different tips, i.e. a topological disagreement carried by tens of thousands of sites in each, not a scarcity effect; consistent with the intra-*Androsace* conflict above.
**Not established.** Separate contribution of entropy / quartet / pair balance (no ablation); stability of the tree between two independent draws (one draw per setting; A → B shows the topology moves with the draw); whether the quartet-informative restriction biases branch lengths (ASC corrects for variable-only ascertainment, not for requiring two states each in two genomes; tree length 22.5 is not interpretable as substitutions/site).
## Open questions
- Per-family $S_f$ without materializing the full matrix: a first pass storing only a per-family summary (reach, $|S_f|$), as `--shannon` already does a full unsampled scan for entropy.
+12
View File
@@ -42,6 +42,18 @@ obikmer:
debug:
cargo build --manifest-path $(CARGO_DIR)/Cargo.toml --bin obikmer
# One-shot converter of indexes built with ptr_hash 1.x (see tools/migrate-mphf).
# Standalone crate, outside the src/ workspace, with its own Cargo.lock/target.
MIGRATE_MPHF_DIR := tools/migrate-mphf
.PHONY: migrate-mphf
migrate-mphf:
cargo build --release --manifest-path $(MIGRATE_MPHF_DIR)/Cargo.toml
.PHONY: clean-migrate-mphf
clean-migrate-mphf:
rm -rf $(MIGRATE_MPHF_DIR)/target
.PHONY: profile-build
profile-build:
cargo build --release --manifest-path $(CARGO_DIR)/Cargo.toml \
+1 -1
View File
@@ -51,7 +51,7 @@ Assuming equal genome lengths, the number of genomes containing at least one occ
$$
P(X \geq r)=\sum_{j=r}^{S}\binom{S}{j}\left(1-e^{-2G/4^k}\right)^j\left(e^{-2G/4^k}\right){S-j}.
P(X \geq r)=\sum_{j=r}^{S}\binom{S}{j}\left(1-e^{-2G/4^k}\right)^j\left(e^{-2G/4^k}\right)^{S-j}.
$$
This sum is the regularized incomplete beta function $I_p(r,\,S-r+1)$, where
+30 -6
View File
@@ -16,7 +16,7 @@ obikmer phylo INDEX [OPTIONS]
| Option | Default | Description |
|---|---|---|
| `--distance` | `jaccard` | See the two tables below for the full list of accepted values |
| `--distance` | none | Compute and write a distance matrix. See the two tables below for the full list of accepted values. No matrix is computed unless this is given, or `--nj`, `--upgma` or `--shared-kmers` is (they need a matrix and then use `jaccard`) |
| `--gamma-shape ALPHA\|auto` | none | Rate-heterogeneity correction, for `snp-*` values that support it (see [Gamma rate heterogeneity](../theory/phylogeny/snp_based/gamma_rate_heterogeneity.md)). Either a fixed $\alpha$ or `auto`/`estimate` to fit it from the data. No effect on the other values; rejected if given together with a value that doesn't support it |
| `--presence-threshold` | `1` | Minimum count for a kmer to be considered present, for `jaccard`/`mash` on a count index |
| `--csv` | off | Write the matrix as plain CSV instead of the default relaxed-PHYLIP format |
@@ -31,7 +31,7 @@ Every value routes to one of two independent computations: **whole-index distanc
### Output
Without `-o`, the matrix goes to stdout in relaxed-PHYLIP format (`n` on the first line, then one `label<TAB>value...` row per genome). With `--csv`, the format is instead a header row `genome,<label1>,<label2>,...` followed by one `<label>,<value1>,<value2>,...` row per genome, 6 decimals. Both formats are symmetric with a zero diagonal, except where noted below.
When a matrix is computed and there is no `-o`, it goes to stdout in relaxed-PHYLIP format (`n` on the first line, then one `label<TAB>value...` row per genome). With `--csv`, the format is instead a header row `genome,<label1>,<label2>,...` followed by one `<label>,<value1>,<value2>,...` row per genome, 6 decimals. Both formats are symmetric with a zero diagonal, except where noted below.
## `--exclude-genome`, `--min-shared-family`
@@ -48,11 +48,12 @@ See [Central-position SNP model](../theory/phylogeny/snp_based/central_snp_model
| Option | Description |
|---|---|
| `--sibling-annex` | Build (or rebuild) the sibling-count/minorant annex — prerequisite for every option in this section, and for a `snp-*` `--distance` value |
| `--sibling-annex` | Build (or rebuild) the sibling-count/minorant annex, then every annex derived from it (entropy, quartet, pair weight, family overlap, under both `--no-ambiguity` conventions). Every option in this section needs the sibling annex, and builds it by itself, under an exclusive lock on the index, if it is missing (as it is after `migrate-mphf`); running `--sibling-annex` first does the same work once, which is the safe way to start several `phylo` runs on the same index at the same time |
| `--sibling-stats` | Write `<prefix>_siblings.csv`: the family-size distribution, per genome and globally |
| `--sibling-hist` | Print the global family-size histogram (1-4 members) only |
| `--family-overlap` | Write `<prefix>_family_overlap.csv`: for every genome pair, how many variable families both genomes carry a call for |
| `--shannon` | Write `<prefix>_entropy.csv`: per-family Shannon entropy, one row per family, full unsampled scan |
| `--shannon-summary` | Write `<prefix>_shannon_summary.csv`: the same scan as `--shannon`, reduced to a distribution (see below) — a few hundred thousand lines at most, whatever the size of the index |
| `--pseudo-alignment` | Write `<prefix>_alignment.fasta`: a SNP-only pseudo-alignment. Requires `--n-sites N\|all` |
### `--sibling-stats`
@@ -81,6 +82,8 @@ See [Central-position SNP model](../theory/phylogeny/snp_based/central_snp_model
| `--no-ambiguity` | off | Treat a genome carrying more than one member of a family as missing data (`?`) instead of an IUPAC ambiguity code |
| `--quartet` | off | Keep only the families that can separate two pairs of genomes from each other: at least two distinct called states, each carried by at least two genomes. Which genomes count as carriers follows `--no-ambiguity` |
| `--pair-balance` | off | Favour the families covering the pairs of genomes that share the fewest families across the index, so that every pair is better represented in the sample. Which genomes count as carriers follows `--no-ambiguity` |
| `--sample-stats` | off | Write three CSV files describing what the draw contains (see "Describing a draw" below) |
| `--sample-plan` | off | Write `<prefix>_sample_plan.csv`: how the draw was spread over the layers of the index |
| `--entropy MU` | off (`1.0` if only `--entropy-sd` is given) | Center of the entropy band to favor when sampling |
| `--entropy-sd SIGMA` | off (`0.5` if only `--entropy` is given) | Width of that band |
@@ -130,6 +133,24 @@ A family carried by fewer than four genomes, or whose genomes all share one stat
The restriction is applied before the draw, from a persisted annex (one bit per variable family, built the first time `--quartet` is used with a given `--no-ambiguity` convention, in one pass shared with the other annexes, and built for both conventions by `--sibling-annex`), then checked exactly on each family drawn, taking `--exclude-genome` into account. The quota given to `--n-sites` is therefore spent only on usable families.
### Describing a draw: `--sample-stats`, `--sample-plan`
Both need a draw to describe (`--pseudo-alignment`, `--sankoff` or a `snp-*` `--distance`) and are computed from the very sites the draw kept, in the same pass, so they cost almost nothing and always agree with the alignment.
`--sample-stats` writes three files:
| File | Content |
|---|---|
| `<prefix>_sample_genomes.csv` | per genome (`genome,excluded,sites_with_call,share_of_sites,informative_sites`): the sampled sites it has a call in, their share of all sampled sites, and the informative sites (two states each carried by at least two genomes) in which its own state is one of those carried by at least two genomes. A genome with almost no calls cannot be placed whatever the tree method |
| `<prefix>_sample_carriers.csv` | per number of carriers (`n_carriers,sites,informative_sites`): how many sampled sites have that many genomes with a call, and how many of them are informative. Shows how much of the sample can resolve a tree at all |
| `<prefix>_sample_overlap.csv` | the sample's own genome × genome matrix, same layout as `--family-overlap`: sampled sites both genomes have a call in. Compare it with the index's matrix to see which pairs the draw under- or over-represents |
A call is any state other than the empty one: absence is never a call, with or without `--free-loss`. Excluded genomes are listed with zeros in the first file and count nowhere.
`--sample-plan` writes `<prefix>_sample_plan.csv`, one line per layer of the index in the order the draw visited them (`partition,layer,eligible,mass,quota,resolved,kept`): the variable families eligible when the draw starts, their total weight (their number, or with `--pair-balance` the sum of their pair weights), the layer's share of the target, the families drawn and resolved, and the ones kept after masking. A low `kept` against `resolved` means masking removes most of what is drawn in that layer; a `quota` summing to less than the target means some layers had nothing eligible. The target and the total kept are logged.
`--shannon-summary` writes `<prefix>_shannon_summary.csv` (`entropy,n_genomes_present,n_genomes_unambiguous,bin_start,n_families`): the number of variable families per entropy (`entropy15` over every carrier, `entropy4` over the single-member genomes only — families with none are left out of its lines), number of genomes carrying the family, number of those without ambiguity, and 0.05-bit entropy bin (`bin_start` is its lower bound). It answers how entropy varies with the number of carriers, or how many carriers `--no-ambiguity` loses, from the same scan as `--shannon` but without one line per family.
### `--pair-balance`: representing every pair of genomes
Two genomes that are distantly related share few families, so a uniform draw covers such a pair with few sites, far fewer than a close pair: the pairs that matter for the deep branches of a tree are the worst represented. `--pair-balance` compensates. Each family is weighted by the mean, over the pairs of genomes carrying it, of 1 divided by the number of families that pair shares across the whole index (the matrix `--family-overlap` writes): a family carried by a rarely-sharing pair weighs more. It is the mean, not the sum, so a family carried by most genomes — in skimming data often a high-copy region — is not favoured merely for covering many pairs.
@@ -177,14 +198,14 @@ TNT and PhyG both write trees with bare numeric leaf labels (`1`, `2`, ..., in t
## Output files
With `-o/--output PREFIX`, the relevant subset of the files below is written. Without `-o`, only the distance matrix is produced, on stdout. All matrices use genome labels as row/column headers, in index order.
With `-o/--output PREFIX`, the relevant subset of the files below is written. Without `-o`, only the distance matrix, if one is computed, is produced, on stdout. All matrices use genome labels as row/column headers, in index order.
### Distance matrix
| File | Written by | Format | Content |
|---|---|---|---|
| `<prefix>_dist.phy` | always, unless `--csv` | relaxed PHYLIP | the `--distance` matrix |
| `<prefix>_dist.csv` | `--csv` | CSV matrix | the `--distance` matrix, 6 decimals |
| `<prefix>_dist.phy` | a distance matrix is computed, unless `--csv` | relaxed PHYLIP | the `--distance` matrix |
| `<prefix>_dist.csv` | a distance matrix is computed, with `--csv` | CSV matrix | the `--distance` matrix, 6 decimals |
| `<prefix>_shared.csv` | `--shared-kmers` | CSV matrix | shared-kmer count per genome pair (integers) |
| `<prefix>_nj.nwk` | `--nj` | Newick | Neighbor-Joining tree |
| `<prefix>_upgma.nwk` | `--upgma` | Newick | UPGMA tree |
@@ -198,6 +219,9 @@ CSV matrix layout (`_dist.csv`, `_shared.csv`, `_family_overlap.csv`): header `g
| `<prefix>_siblings.csv` | `--sibling-stats` | CSV table | family-size distribution, per genome and global |
| `<prefix>_family_overlap.csv` | `--family-overlap` | CSV matrix | variable families both genomes of a pair carry a call for |
| `<prefix>_entropy.csv` | `--shannon` | CSV table | per-family Shannon entropy, see "Sampling at scale" above |
| `<prefix>_shannon_summary.csv` | `--shannon-summary` | CSV table | families per entropy bin, number of carriers and number of single-member carriers |
| `<prefix>_sample_genomes.csv`, `<prefix>_sample_carriers.csv`, `<prefix>_sample_overlap.csv` | `--sample-stats` | CSV | what the draw contains: per genome, per number of carriers, per pair of genomes |
| `<prefix>_sample_plan.csv` | `--sample-plan` | CSV table | per layer: eligible families, weight, quota, drawn, kept |
| `<prefix>_alignment.fasta` | `--pseudo-alignment` | FASTA | SNP-only pseudo-alignment, IUPAC-coded |
### Sankoff calibration and exports
+152
View File
@@ -0,0 +1,152 @@
#!/usr/bin/env bash
# Usage: bench_mphf.sh LABEL
#
# Measures the obikmer commands that sit on MPHF lookups (see
# DevDocMD/implementation/mphf_batch_queries.md for the site inventory), on the
# 20-genome benchmark data already built under run/ (`gmake all` first).
# Nothing under run/ other than run/bench_mphf/ is written.
#
# Env:
# BINARY obikmer binary to measure (default ../src/target/release/obikmer)
# REPS repetitions per step (default 3)
# CPU_MAX passed as --cpu-max when set (default: all cores)
#
# Output: run/bench_mphf/LABEL/results.csv (label,step,rep,wall_s,maxrss_b)
# run/bench_mphf/LABEL/checksums.txt (order-independent output digests,
# to be diffed across LABELs: a speed-up must not change any of them)
set -euo pipefail
LABEL="$1"
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
RUN_DIR="${SCRIPT_DIR}/run"
BINARY="${BINARY:-${SCRIPT_DIR}/../src/target/release/obikmer}"
REPS="${REPS:-3}"
OUT="${RUN_DIR}/bench_mphf/${LABEL}"
CPU=()
[[ -n "${CPU_MAX:-}" ]] && CPU=(--cpu-max "${CPU_MAX}")
rm -rf "${OUT}"; mkdir -p "${OUT}"
RESULTS="${OUT}/results.csv"
SUMS="${OUT}/checksums.txt"
echo "label,step,rep,wall_s,maxrss_b" >"${RESULTS}"
: >"${SUMS}"
# timed STEP REP CMD... — wall time + peak RSS via BSD time(1); stdout of CMD
# goes to $OUT/last.stdout, stderr to $OUT/last.stderr.
timed() {
local step="$1" rep="$2"; shift 2
local tf="${OUT}/time.tmp"
/usr/bin/time -l -o "${tf}" "$@" >"${OUT}/last.stdout" 2>"${OUT}/last.stderr" \
|| { echo "FAILED: $step: $*" >&2; cat "${OUT}/last.stderr" >&2; exit 1; }
local wall rss
wall=$(awk '/ real /{print $1}' "${tf}")
rss=$(awk '/maximum resident set size/{print $1}' "${tf}")
echo "${LABEL},${step},${rep},${wall},${rss}" >>"${RESULTS}"
# per-stage wall times from the Reporter log lines ("done stage=X wall_secs=Y")
sed $'s/\x1b\\[[0-9;]*m//g' "${OUT}/last.stderr" \
| awk -v L="${LABEL}" -v S="${step}" -v R="${rep}" \
'/ done stage=/{ for(i=1;i<=NF;i++){ if($i ~ /^stage=/) st=substr($i,7); if($i ~ /^wall_secs=/) w=substr($i,11) } print L","S":"st","R","w",0" }' \
>>"${RESULTS}"
printf ' %-28s rep %s %8s s\n' "${step}" "${rep}" "${wall}" >&2
}
digest() { LC_ALL=C sort "$1" | md5 -q; }
# ── 0. data: the indexes under run/ may have been built by an older
# MPHF generation (ptr_hash 1.x, unreadable by current binaries), so the
# bench builds its own, once, from the simulated reads. Shared by all LABELs;
# delete run/bench_mphf/data to force a rebuild.
DATA="${RUN_DIR}/bench_mphf/data"
if [[ ! -f "${DATA}/.ready" ]]; then
echo "preparing data with ${BINARY}" >&2
mkdir -p "${DATA}/presence" "${DATA}/count"
for d in "${RUN_DIR}"/simulated_data/*/*/; do
strain="$(basename "$d")"; species="$(basename "$(dirname "$d")")"
sp="${species}--${strain}"
for kind in presence count; do
extra=(); [[ $kind == count ]] && extra=(--with-counts)
"${BINARY}" "${CPU[@]}" index --output "${DATA}/${kind}/${sp}" --force \
--theta 0 "${extra[@]}" --label "${sp}" --meta "species=${species}" \
"$d/reads_R1.fastq.gz" "$d/reads_R2.fastq.gz" 2>/dev/null
done
done
"${BINARY}" "${CPU[@]}" merge --output "${DATA}/global_presence" --force \
--force-presence "${DATA}"/presence/*/ 2>/dev/null
"${BINARY}" "${CPU[@]}" merge --output "${DATA}/global_count" --force \
"${DATA}"/count/*/ 2>/dev/null
touch "${DATA}/.ready"
fi
sources=("${DATA}"/presence/*/)
# ── 1. index: build pass 2 (mphf.index per kmer + evidence writes) ──────────
for sp in Escherichia_coli--K-12_MG1655 Candidozyma_auris--GCF_003013715.1_ASM301371v2; do
species="${sp%%--*}"; strain="${sp#*--}"
reads="${RUN_DIR}/simulated_data/${species}/${strain}"
for rep in $(seq "${REPS}"); do
timed "index_${species}" "${rep}" "${BINARY}" "${CPU[@]}" index \
--output "${OUT}/idx_${species}" --force --theta 0 --label "${sp}" \
"${reads}/reads_R1.fastq.gz" "${reads}/reads_R2.fastq.gz"
done
done
# ── 2. merge: find_batch_grouped, build_graph filter, hash_batch, index_batch ─
for rep in $(seq "${REPS}"); do
timed "merge_20_presence" "${rep}" "${BINARY}" "${CPU[@]}" merge \
--output "${OUT}/merged" --force --force-presence "${sources[@]}"
done
"${BINARY}" "${CPU[@]}" dump "${OUT}/merged" >"${OUT}/merged.dump"
echo "merged_dump $(digest "${OUT}/merged.dump")" >>"${SUMS}"
rm -f "${OUT}/merged.dump"
# ── 3. query: query stage 1 (find per kmer, layer by layer) ─────────────────
for kind in presence count; do
for sp in Escherichia_coli/K-12_MG1655 Saccharolobus_islandicus/M.16.4; do
reads="${RUN_DIR}/query_data/${sp}"
for rep in $(seq "${REPS}"); do
timed "query_${kind}_${sp%%/*}" "${rep}" "${BINARY}" "${CPU[@]}" query \
--count-missing "${DATA}/global_${kind}" \
"${reads}/reads_R1.fastq.gz" "${reads}/reads_R2.fastq.gz"
done
echo "query_${kind}_${sp%%/*} $(digest "${OUT}/last.stdout")" >>"${SUMS}"
done
done
# ── 4. filter: iter_layer_kmers → hash_batch + fill_sub_matrix ──────────────
for rep in $(seq "${REPS}"); do
timed "filter_species" "${rep}" "${BINARY}" "${CPU[@]}" filter \
--output "${OUT}/filtered" --force --ingroup "species=Escherichia_coli" \
--outgroup all --min-frac 0.5 --max-frac 1.0 --max-outgroup-count 0 \
"${DATA}/global_presence"
done
"${BINARY}" "${CPU[@]}" dump "${OUT}/filtered" >"${OUT}/filtered.dump"
echo "filtered_dump $(digest "${OUT}/filtered.dump")" >>"${SUMS}"
rm -f "${OUT}/filtered.dump"
# ── 5. convert: build_exact_evidence / build_approx_evidence ────────────────
# On an APFS clone (cp -c) so run/global_index_presence is untouched.
for rep in $(seq "${REPS}"); do
rm -rf "${OUT}/conv"; cp -cR "${DATA}/global_presence" "${OUT}/conv"
timed "convert_to_approx8" "${rep}" "${BINARY}" "${CPU[@]}" convert \
--approx-evidence 8 "${OUT}/conv"
timed "convert_to_exact" "${rep}" "${BINARY}" "${CPU[@]}" convert \
--exact-evidence "${OUT}/conv"
done
"${BINARY}" "${CPU[@]}" dump "${OUT}/conv" >"${OUT}/conv.dump"
echo "convert_roundtrip_dump $(digest "${OUT}/conv.dump")" >>"${SUMS}"
rm -rf "${OUT}/conv" "${OUT}/conv.dump"
# ── 6. phylo: sibling annex (annex.rs: find_in_partition per variant) and
# its consumer (family_scan.rs: hash_batch / find_slot) ──────────────────
for rep in $(seq "${REPS}"); do
rm -rf "${OUT}/phy"; cp -cR "${DATA}/global_presence" "${OUT}/phy"
timed "phylo_sibling_annex" "${rep}" "${BINARY}" "${CPU[@]}" phylo \
--sibling-annex "${OUT}/phy"
timed "phylo_snp_distance" "${rep}" "${BINARY}" "${CPU[@]}" phylo \
--output "${OUT}/snp" --distance snp-jc --csv "${OUT}/phy"
done
echo "phylo_snp_dist $(digest "${OUT}/snp_dist.csv")" >>"${SUMS}"
rm -rf "${OUT}/phy" "${OUT}/idx_"* "${OUT}/merged" "${OUT}/filtered" \
"${OUT}/last.stdout" "${OUT}/time.tmp"
echo "results: ${RESULTS}" >&2
echo "checksums: ${SUMS}" >&2
+14
View File
@@ -0,0 +1,14 @@
#!/usr/bin/env python3
"""Usage: summarize_bench.py LABEL [LABEL...] — median wall per step, side by side."""
import csv, statistics, sys, os
base = os.path.join(os.path.dirname(os.path.abspath(__file__)), "run", "bench_mphf")
data = {}
for lab in sys.argv[1:]:
for r in csv.DictReader(open(f"{base}/{lab}/results.csv")):
data.setdefault(r["step"], {}).setdefault(lab, []).append(float(r["wall_s"]))
print(f"{'step':32}" + "".join(f"{l:>14}" for l in sys.argv[1:]) + (" ratio" if len(sys.argv) > 2 else ""))
for step, d in data.items():
med = [statistics.median(d[l]) for l in sys.argv[1:]]
line = f"{step:32}" + "".join(f"{m:14.2f}" for m in med)
if len(med) > 1 and med[0]: line += f" {med[-1]/med[0]:.3f}"
print(line)
+2
View File
@@ -48,6 +48,7 @@ nav:
- obipipeline library: implementation/obipipeline.md
- On-disk storage: implementation/storage.md
- MPHF selection: implementation/mphf.md
- "MPHF batch queries (inventory)": implementation/mphf_batch_queries.md
- Unitig evidence encoding: implementation/unitig_evidence.md
- Evidence elimination (discussion): implementation/evidence_elimination.md
- obikindex layer tier: implementation/layer_tier.md
@@ -64,6 +65,7 @@ nav:
- Sequences: architecture/sequences/invariant.md
- Kmer index: architecture/index_architecture.md
- Sibling annex: architecture/siblings.md
- Hot resume and memoization: architecture/hot_resume_memoization.md
- NUMA-aware worker pools: architecture/numa_worker_pools.md
- NUMA-aware partition runner: architecture/numa_partition_runner.md
+24
View File
@@ -0,0 +1,24 @@
#!/usr/bin/env python3
# check_unitig_chunks.py GLOB... — report unitigs.bin chunks holding more than 128 k-mers.
# Indexes built before MAX_KMERS_PER_CHUNK went from 256 to 128 (2026-10-15) could hold chunks whose
# rank does not fit the 7-bit evidence field; the k-mers beyond rank 127 of such a chunk then fail
# exact verification. Example: scripts/check_unitig_chunks.py "<index>/partitions/part_*/index/layer_*/unitigs.bin"
# (k is hard-coded below — edit it for an index built with another k.)
import sys,glob,mmap
k=31
def scan(path):
with open(path,'rb') as f:
m=mmap.mmap(f.fileno(),0,access=mmap.ACCESS_READ)
o=0; n=len(m); mx=0; over=0; chunks=0
while o<n:
c=m[o]+1; chunks+=1
if c>mx: mx=c
if c>128: over+=1
o+=1+(m[o]+k+3)//4
return mx,over,chunks
tot_over=0
for pat in sys.argv[1:]:
for f in sorted(glob.glob(pat)):
mx,over,ch=scan(f); tot_over+=over
print(f"{f.split('partitions/')[1]}: chunks={ch} max_kmers={mx} chunks>128={over}")
print("TOTAL chunks over 128 kmers:",tot_over)
+152 -187
View File
@@ -8,19 +8,6 @@ version = "2.0.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "320119579fcad9c21884f5c4861d16174d0e06250625266f50fe6898340abefa"
[[package]]
name = "ahash"
version = "0.8.12"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5a15f179cd60c4584b8a8c596927aadc462e27f2ca70c04e0071964a73ba7a75"
dependencies = [
"cfg-if",
"getrandom 0.3.4",
"once_cell",
"version_check",
"zerocopy",
]
[[package]]
name = "aho-corasick"
version = "1.1.4"
@@ -30,6 +17,12 @@ dependencies = [
"memchr",
]
[[package]]
name = "aliasable"
version = "0.1.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "250f629c0161ad8107cf89319e990051fae62832fd343083bea452d93e2205fd"
[[package]]
name = "allocator-api2"
version = "0.2.21"
@@ -167,12 +160,6 @@ version = "0.6.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "349f9b6a179ed607305526ca489b34ad0a41aed5f7980fa90eb03160b69598fb"
[[package]]
name = "bitflags"
version = "1.3.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "bef38d45163c2f1dde094a7dfd33ccf595c92905c8f8f4fdc18d06fb1037718a"
[[package]]
name = "bitflags"
version = "2.11.1"
@@ -191,6 +178,15 @@ dependencies = [
"wyz",
]
[[package]]
name = "block-buffer"
version = "0.10.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3078c7629b62d3f0439517fa394996acacc5cbc91c5a20d8c658e77abd503a71"
dependencies = [
"generic-array",
]
[[package]]
name = "block-buffer"
version = "0.12.1"
@@ -268,11 +264,11 @@ dependencies = [
[[package]]
name = "cacheline-ef"
version = "1.1.0"
version = "2.0.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "af737c6c59cb018ecbe6472cbdf86d39c59d78252febfe311953a991b6e4ed85"
checksum = "6e701af8ea9f58c4ba18793a5d5dffd791cda99bf766976c59c583939d180fd5"
dependencies = [
"common_traits 0.11.4",
"common_traits",
"epserde",
"mem_dbg",
]
@@ -312,6 +308,12 @@ version = "1.0.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9330f8b2ff13f34540b44e946ef35111825727b38d33286ef986142615121801"
[[package]]
name = "cfg_aliases"
version = "0.2.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f079e83a288787bcd14a6aea84cee5c87a67c5a3e660c30f557a3d24761b3527"
[[package]]
name = "chacha20"
version = "0.10.1"
@@ -319,7 +321,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d524456ba66e72eb8b115ff89e01e497f8e6d11d78b70b1aa13c0fbd97540a81"
dependencies = [
"cfg-if",
"cpufeatures",
"cpufeatures 0.3.0",
"rand_core 0.10.1",
]
@@ -433,17 +435,6 @@ dependencies = [
"memchr",
]
[[package]]
name = "common_traits"
version = "0.11.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "fda9ae1f26adcae83adb2e92f69cf59421f2a277a942f49f8e59f2fcbd7cf062"
dependencies = [
"anyhow",
"half",
"impl-tools 0.10.3",
]
[[package]]
name = "common_traits"
version = "0.13.0"
@@ -452,7 +443,7 @@ checksum = "d8e0d7d0c5977b1e87f759d406740c38227125f374f54d81bc197e877f2c5f26"
dependencies = [
"anyhow",
"half",
"impl-tools 0.12.0",
"impl-tools",
]
[[package]]
@@ -482,6 +473,15 @@ dependencies = [
"unicode-segmentation",
]
[[package]]
name = "cpufeatures"
version = "0.2.17"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "59ed5838eebb26a2bb2e58f6d5b5316989ae9d08bab10e0e6d103e656d1b0280"
dependencies = [
"libc",
]
[[package]]
name = "cpufeatures"
version = "0.3.0"
@@ -592,6 +592,16 @@ version = "0.2.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "460fbee9c2c2f33933d720630a6a0bac33ba7053db5344fac858d4b8952d77d5"
[[package]]
name = "crypto-common"
version = "0.1.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "78c8292055d1c1df0cce5d180393dc8cce0abec0a7102adb6c7b1eef6016d60a"
dependencies = [
"generic-array",
"typenum",
]
[[package]]
name = "crypto-common"
version = "0.2.2"
@@ -645,15 +655,25 @@ dependencies = [
"unicode-xid",
]
[[package]]
name = "digest"
version = "0.10.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9ed9a281f7bc9b7576e61468ba615a66a5c8cfdff42420a70aa82701a3b1e292"
dependencies = [
"block-buffer 0.10.4",
"crypto-common 0.1.7",
]
[[package]]
name = "digest"
version = "0.11.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f1dd6dbb5841937940781866fa1281a1ff7bd3bf827091440879f9994983d5c2"
dependencies = [
"block-buffer",
"block-buffer 0.12.1",
"const-oid",
"crypto-common",
"crypto-common 0.2.2",
]
[[package]]
@@ -662,7 +682,7 @@ version = "0.3.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "1e0e367e4e7da84520dedcac1901e4da967309406d1e51017ae1abfb97adbd38"
dependencies = [
"bitflags 2.11.1",
"bitflags",
"objc2",
]
@@ -698,27 +718,27 @@ dependencies = [
[[package]]
name = "epserde"
version = "0.8.0"
version = "0.13.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "c40d342ff20a2ce62d9a85ce406e672dfa137f902ac9670034533184f1533976"
checksum = "d6cd9f3212ce43e7b594a974c6e8cef4e11c6473d76b7b97cb01c239130789eb"
dependencies = [
"aliasable",
"anyhow",
"bitflags 2.11.1",
"common_traits 0.11.4",
"bitflags",
"epserde-derive",
"maligned",
"maybe-dangling",
"mem_dbg",
"mmap-rs",
"sealed",
"sha2",
"thiserror 2.0.18",
"xxhash-rust",
]
[[package]]
name = "epserde-derive"
version = "0.8.0"
version = "0.13.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ac80cc78b69765703f48ad93f33b8919cf5d907cda7459ad6ba2919cbbe605dd"
checksum = "d73e3083270b2ff144c990e26580374fdfd512f61dd5a12da9b5bd749a7585e0"
dependencies = [
"proc-macro2",
"quote",
@@ -806,6 +826,16 @@ dependencies = [
"byteorder",
]
[[package]]
name = "generic-array"
version = "0.14.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "85649ca51fd72272d7821adaf274ad91c288277713d9c18820d8499a7ff69e9a"
dependencies = [
"typenum",
"version_check",
]
[[package]]
name = "getrandom"
version = "0.2.17"
@@ -852,6 +882,17 @@ dependencies = [
"zerocopy",
]
[[package]]
name = "hashbrown"
version = "0.16.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "841d1cc9bed7f9236f321df977030373f4a4163ae1a7dbfe1a51a2c1a51d9100"
dependencies = [
"allocator-api2",
"equivalent",
"foldhash",
]
[[package]]
name = "hashbrown"
version = "0.17.1"
@@ -899,7 +940,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4c2e65a48d3b300843ac84a2fe8e166bb5a5b00f30054593bcee8157e4b465fd"
dependencies = [
"arrayvec",
"bitflags 2.11.1",
"bitflags",
"derive_more",
"errno",
"hwlocality-sys",
@@ -935,18 +976,6 @@ dependencies = [
"typenum",
]
[[package]]
name = "impl-tools"
version = "0.10.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0ae95c9095c2f1126d7db785955c73cdc5fc33e7c3fa911bd4a42931672029a7"
dependencies = [
"autocfg",
"impl-tools-lib 0.11.4",
"proc-macro-error2",
"syn 2.0.117",
]
[[package]]
name = "impl-tools"
version = "0.12.0"
@@ -954,23 +983,11 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "06c8567c75b167cdda5ecee024da9cc9581d41bd52460e2e0e768967972aeef1"
dependencies = [
"autocfg",
"impl-tools-lib 0.12.0",
"impl-tools-lib",
"proc-macro-error3",
"syn 2.0.117",
]
[[package]]
name = "impl-tools-lib"
version = "0.11.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ab699036df31c1f7d3561bfa6e9cb9bc3bb0fd2e2cd9bf121c31cb961d049ddf"
dependencies = [
"proc-macro-error2",
"proc-macro2",
"quote",
"syn 2.0.117",
]
[[package]]
name = "impl-tools-lib"
version = "0.12.0"
@@ -990,7 +1007,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d466e9454f08e4a911e14806c24e16fba1b4c121d1ea474396f396069cf949d9"
dependencies = [
"equivalent",
"hashbrown",
"hashbrown 0.17.1",
]
[[package]]
@@ -1043,9 +1060,9 @@ dependencies = [
[[package]]
name = "itertools"
version = "0.14.0"
version = "0.15.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2b192c782037fadd9cfa75548310488aabdbf3d2da73885b31bd0abd03351285"
checksum = "8b4baf93f58d4425749ca49a51c50ebab072c5df6994d08fed93541c331481dc"
dependencies = [
"either",
]
@@ -1083,7 +1100,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9e24a010dd405bd7ed803e5253182815b41bf2e6a80cc3bfc066658e03a198aa"
dependencies = [
"cfg-if",
"cpufeatures",
"cpufeatures 0.3.0",
]
[[package]]
@@ -1157,7 +1174,7 @@ version = "0.18.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5d2f2f9b4ba7e6b24d95e7e899329d35be83bcded72c8540cdd5368932d1d90a"
dependencies = [
"hashbrown",
"hashbrown 0.17.1",
]
[[package]]
@@ -1169,12 +1186,6 @@ dependencies = [
"libc",
]
[[package]]
name = "maligned"
version = "0.2.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7e88c3cbe8288f77f293e48a28b3232e3defd203a6d839fa7f68ea4329e83464"
[[package]]
name = "matchers"
version = "0.2.0"
@@ -1195,22 +1206,30 @@ dependencies = [
]
[[package]]
name = "mem_dbg"
version = "0.3.4"
name = "maybe-dangling"
version = "0.1.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "728cc9dc97593cd22f7bc81fbef70a2d391d7a9a855e7d658b653318124a6cf0"
checksum = "59dbb09ed53f8e4f314e353dc6c1853ae5b4c480a668a422657804a544ea9f65"
[[package]]
name = "mem_dbg"
version = "0.4.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b48a1086c746f4ee6ca5cb0acf856a14709bc4d2d20e03db150a12ddf2269e6d"
dependencies = [
"bitflags 2.11.1",
"maligned",
"aliasable",
"bitflags",
"hashbrown 0.16.1",
"maybe-dangling",
"mem_dbg-derive",
"mmap-rs",
]
[[package]]
name = "mem_dbg-derive"
version = "0.2.1"
version = "0.3.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d84f40c93b0508d5565db79a814d02d5b2545967205ce44be211592aafa34d6c"
checksum = "eb910efe8da52f13da727170e352e50a1764579a6fb1065d00d9556da19c79ac"
dependencies = [
"proc-macro2",
"quote",
@@ -1232,15 +1251,6 @@ dependencies = [
"libc",
]
[[package]]
name = "memoffset"
version = "0.7.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5de893c32cde5f383baa4c04c5d6dbdd735cfd4a794b0debdb2bb1b421da5ff4"
dependencies = [
"autocfg",
]
[[package]]
name = "miniz_oxide"
version = "0.8.9"
@@ -1253,17 +1263,17 @@ dependencies = [
[[package]]
name = "mmap-rs"
version = "0.6.1"
version = "0.7.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "86968d85441db75203c34deefd0c88032f275aaa85cee19a1dcfff6ae9df56da"
checksum = "4ecce9d566cb9234ae3db9e249c8b55665feaaf32b0859ff1e27e310d2beb3d8"
dependencies = [
"bitflags 1.3.2",
"bitflags",
"combine",
"libc",
"mach2",
"nix",
"sysctl",
"thiserror 1.0.69",
"thiserror 2.0.18",
"widestring",
"windows 0.48.0",
]
@@ -1326,15 +1336,14 @@ dependencies = [
[[package]]
name = "nix"
version = "0.26.4"
version = "0.30.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "598beaf3cc6fdd9a5dfb1630c2800c7acd31df7aaf0f565796fba2b53ca1af1b"
checksum = "74523f3a35e05aba87a1d978330aef40f67b0304ac79c1c00b294c9830543db6"
dependencies = [
"bitflags 1.3.2",
"bitflags",
"cfg-if",
"cfg_aliases",
"libc",
"memoffset",
"pin-utils",
]
[[package]]
@@ -1386,7 +1395,7 @@ dependencies = [
name = "obicompactvec"
version = "0.1.0"
dependencies = [
"common_traits 0.13.0",
"common_traits",
"memmap2",
"ndarray",
"rayon",
@@ -1397,14 +1406,12 @@ dependencies = [
name = "obidebruinj"
version = "0.1.0"
dependencies = [
"ahash",
"crossbeam-channel",
"hashbrown",
"hashbrown 0.17.1",
"obifastwrite",
"obikseq",
"rayon",
"tracing",
"xxhash-rust",
]
[[package]]
@@ -1506,7 +1513,6 @@ dependencies = [
name = "obikindexer"
version = "0.1.0"
dependencies = [
"cacheline-ef",
"epserde",
"memmap2",
"niffler",
@@ -1521,7 +1527,6 @@ dependencies = [
"obiskbuilder",
"obiskio",
"obisys",
"ptr_hash",
"rayon",
"serde",
"serde_json",
@@ -1532,7 +1537,7 @@ dependencies = [
[[package]]
name = "obikmer"
version = "1.3.6"
version = "1.4.1"
dependencies = [
"clap",
"csv",
@@ -1773,7 +1778,7 @@ version = "0.3.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2a180dd8642fa45cdb7dd721cd4c11b1cadd4929ce112ebd8b9f5803cc79d536"
dependencies = [
"bitflags 2.11.1",
"bitflags",
"dispatch2",
"objc2",
]
@@ -1790,7 +1795,7 @@ version = "0.3.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e3e0adef53c21f888deb4fa59fc59f7eb17404926ee8a6f59f5df0fd7f9f3272"
dependencies = [
"bitflags 2.11.1",
"bitflags",
"objc2",
]
@@ -1884,12 +1889,6 @@ version = "0.2.17"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a89322df9ebe1c1578d689c92318e070967d1042b512afbe49518723f4e6d5cd"
[[package]]
name = "pin-utils"
version = "0.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8b870d8c151b6f2fb93e84a13146138f05d02ed11c7e7c54f8826aaaf7c9f184"
[[package]]
name = "pkg-config"
version = "0.3.33"
@@ -1949,14 +1948,10 @@ dependencies = [
]
[[package]]
name = "proc-macro-error-attr2"
version = "2.0.0"
name = "prefetch-index"
version = "0.2.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "96de42df36bb9bba5542fe9f1a054b8cc87e172759a1868aa05c1f3acc89dfc5"
dependencies = [
"proc-macro2",
"quote",
]
checksum = "9057806a8d77d67bccdc0f542db43737a6f19ada3efab2adc63277feea27310f"
[[package]]
name = "proc-macro-error-attr3"
@@ -1968,17 +1963,6 @@ dependencies = [
"quote",
]
[[package]]
name = "proc-macro-error2"
version = "2.0.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "11ec05c52be0a07b08061f7dd003e7d7092e0472bc731b4af7bb1ef876109802"
dependencies = [
"proc-macro-error-attr2",
"proc-macro2",
"quote",
]
[[package]]
name = "proc-macro-error3"
version = "3.1.0"
@@ -2001,30 +1985,25 @@ dependencies = [
[[package]]
name = "ptr_hash"
version = "1.1.0"
version = "2.1.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4b4e4fb9c4c2ba3e5b060f53ef46afd3de37345b08e3ec0f2c65e0ca1d57ccbd"
checksum = "b22cf1be774892680e056d3387db0dcaa4fd46964b270671ef661e127186b969"
dependencies = [
"anyhow",
"bitvec",
"cacheline-ef",
"clap",
"colored",
"common_traits 0.11.4",
"epserde",
"epserde-derive",
"fastrand",
"fxhash",
"itertools 0.14.0",
"lazy_static",
"itertools 0.15.0",
"log",
"mem_dbg",
"rand 0.9.4",
"rand_chacha 0.9.0",
"prefetch-index",
"rand 0.10.2",
"rand_chacha 0.10.0",
"rayon",
"rdst",
"rustc-hash",
"serde",
"sucds",
"tempfile",
"xxhash-rust",
@@ -2097,16 +2076,6 @@ dependencies = [
"rand_core 0.6.4",
]
[[package]]
name = "rand"
version = "0.9.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "44c5af06bb1b7d3216d91932aed5265164bf384dc89cd6ba05cf59a35f5f76ea"
dependencies = [
"rand_chacha 0.9.0",
"rand_core 0.9.5",
]
[[package]]
name = "rand"
version = "0.10.2"
@@ -2130,12 +2099,12 @@ dependencies = [
[[package]]
name = "rand_chacha"
version = "0.9.0"
version = "0.10.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d3022b5f1df60f26e1ffddd6c66e8aa15de382ae63b3a0c1bfc0e4d3e3f325cb"
checksum = "3e6af7f3e25ded52c41df4e0b1af2d047e45896c2f3281792ed68a1c243daedb"
dependencies = [
"ppv-lite86",
"rand_core 0.9.5",
"rand_core 0.10.1",
]
[[package]]
@@ -2147,15 +2116,6 @@ dependencies = [
"getrandom 0.2.17",
]
[[package]]
name = "rand_core"
version = "0.9.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "76afc826de14238e6e8c374ddcc1fa19e374fd8dd986b0d2af0d02377261d83c"
dependencies = [
"getrandom 0.3.4",
]
[[package]]
name = "rand_core"
version = "0.10.1"
@@ -2215,7 +2175,7 @@ version = "0.5.18"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ed2bf2547551a7053d6fdfafda3f938979645c44812fbfcda098faae3f1a362d"
dependencies = [
"bitflags 2.11.1",
"bitflags",
]
[[package]]
@@ -2278,7 +2238,7 @@ checksum = "d9776093b7ca170454ab1406954f7b7d97a57c51dc6c0642957fb2ef25c2d399"
dependencies = [
"bytecheck",
"bytes",
"hashbrown",
"hashbrown 0.17.1",
"indexmap",
"munge",
"ptr_meta",
@@ -2300,12 +2260,6 @@ dependencies = [
"syn 3.0.3",
]
[[package]]
name = "rustc-hash"
version = "2.1.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "94300abf3f1ae2e2b8ffb7b58043de3d399c73fa6f4b73826402a5c457614dbe"
[[package]]
name = "rustc_version"
version = "0.4.1"
@@ -2321,7 +2275,7 @@ version = "1.1.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b6fe4565b9518b83ef4f91bb47ce29620ca828bd32cb7e408f0062e9930ba190"
dependencies = [
"bitflags 2.11.1",
"bitflags",
"errno",
"libc",
"linux-raw-sys",
@@ -2463,13 +2417,24 @@ dependencies = [
"unsafe-libyaml",
]
[[package]]
name = "sha2"
version = "0.10.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a7507d819769d01a365ab707794a4084392c824f54a7a6a7862f8c3d0892b283"
dependencies = [
"cfg-if",
"cpufeatures 0.2.17",
"digest 0.10.7",
]
[[package]]
name = "sha3"
version = "0.11.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "be176f1a57ce4e3d31c1a166222d9768de5954f811601fb7ca06fc8203905ce1"
dependencies = [
"digest",
"digest 0.11.3",
"keccak",
]
@@ -2592,11 +2557,11 @@ dependencies = [
[[package]]
name = "sysctl"
version = "0.5.5"
version = "0.6.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ec7dddc5f0fee506baf8b9fdb989e242f17e4b11c61dfbb0635b705217199eea"
checksum = "01198a2debb237c62b6826ec7081082d951f46dbb64b0e8c7649a452230d1dfc"
dependencies = [
"bitflags 2.11.1",
"bitflags",
"byteorder",
"enum-as-inner",
"libc",
+43 -5
View File
@@ -25,6 +25,7 @@ use std::fs;
use std::io;
use std::path::{Path, PathBuf};
use memmap2::Mmap;
use ndarray::{Array1, Array2};
use crate::traits::{BitPartials, ColumnWeights};
@@ -151,7 +152,9 @@ pub struct PersistentSparseBitMatrix {
singleton: PersistentFixedIntVec,
multi: PersistentFixedIntVec,
dict_offsets: EliasFano,
dict_values: Vec<u8>,
/// `dict_values.bin`, mapped (never copied to the heap); `None` when the
/// file is empty — no multi-genome set — since an empty file cannot be mapped.
dict_values: Option<Mmap>,
/// Row count (k-mer slots).
n: usize,
/// Column count (genomes).
@@ -167,7 +170,12 @@ impl PersistentSparseBitMatrix {
let singleton = PersistentFixedIntVec::open(&singleton_path(dir))?;
let multi = PersistentFixedIntVec::open(&multi_path(dir))?;
let dict_offsets = EliasFano::open(&dict_offsets_base(dir))?;
let dict_values = fs::read(dict_values_path(dir))?;
let dict_file = fs::File::open(dict_values_path(dir))?;
let dict_values = if dict_file.metadata()?.len() == 0 {
None
} else {
Some(unsafe { Mmap::map(&dict_file)? })
};
Ok(Self {
is_multi, singleton, multi, dict_offsets, dict_values,
n: meta.n, n_cols: meta.n_cols, n_distinct_multi: meta.n_distinct_multi,
@@ -179,6 +187,11 @@ impl PersistentSparseBitMatrix {
#[inline]
pub fn n_cols(&self) -> usize { self.n_cols }
#[inline]
fn dict_bytes(&self) -> &[u8] {
self.dict_values.as_deref().unwrap_or(&[])
}
/// Byte range of `dict_id`'s varint-encoded genome list within
/// `dict_values`.
#[inline]
@@ -187,7 +200,7 @@ impl PersistentSparseBitMatrix {
let end = if dict_id + 1 < self.n_distinct_multi {
self.dict_offsets.get(dict_id + 1) as usize
} else {
self.dict_values.len()
self.dict_bytes().len()
};
(start, end)
}
@@ -232,7 +245,7 @@ impl PersistentSparseBitMatrix {
let (start, end) = self.dict_entry_range(dict_id);
let mut p = start;
while p < end {
f(read_varint(&self.dict_values, &mut p) as usize);
f(read_varint(self.dict_bytes(), &mut p) as usize);
}
} else {
let pos = self.is_multi.rank0(slot) as usize;
@@ -287,7 +300,7 @@ impl PersistentSparseBitMatrix {
present.clear();
let mut p = start;
while p < end {
present.push(read_varint(&self.dict_values, &mut p) as usize);
present.push(read_varint(self.dict_bytes(), &mut p) as usize);
}
for &g in &present {
counts[g] += w;
@@ -514,6 +527,31 @@ impl PersistentSparseBitMatrixBuilder {
}
Ok(builder)
}
/// Builds the row-permuted copy of `src` into `dir`: row `s` of the
/// result is row `new_to_old[s]` of `src`. `new_to_old` must have
/// `src.n()` entries and be a permutation of `0..src.n()` — the caller
/// proves that, this only reads `src` at those slots.
///
/// Needed whenever the slot numbering of a layer changes (its MPHF is
/// rebuilt) while the data attached to each k-mer must follow it.
/// Rows are read at arbitrary slots of `src`, so convert a dense source
/// to sparse first (`build_from_dense`) rather than permuting it
/// cell by cell.
pub fn build_permuted(
src: &PersistentSparseBitMatrix,
new_to_old: &[u32],
dir: &Path,
) -> io::Result<Self> {
let mut builder = Self::new(new_to_old.len(), src.n_cols(), dir)?;
let mut genomes: Vec<u32> = Vec::new();
for &old in new_to_old {
genomes.clear();
src.for_each_genome_in_row(old as usize, |g| genomes.push(g as u32));
builder.push_row(&genomes);
}
Ok(builder)
}
}
// ── Trait impls ───────────────────────────────────────────────────────────────
+6 -5
View File
@@ -4,7 +4,7 @@ use std::path::{Path, PathBuf};
use memmap2::{Mmap, MmapMut};
use crate::mmap_file::{create_mmap_file, open_mmap_file};
use crate::mmap_file::{building_path, create_mmap_file, open_mmap_file, publish_built};
use crate::reader::PersistentCompactIntVec;
use crate::views::{BitSliceIter, BitSliceView, IntSliceView};
@@ -206,7 +206,7 @@ impl PersistentBitVecBuilder {
.write(true)
.create(true)
.truncate(true)
.open(path)?;
.open(building_path(path))?;
file.write_all(&MAGIC)?;
file.write_all(&[0u8; 4])?;
file.write_all(&(n as u64).to_le_bytes())?;
@@ -229,8 +229,8 @@ impl PersistentBitVecBuilder {
}
pub fn build_from(source: &PersistentBitVec, path: &Path) -> io::Result<Self> {
fs::copy(source.path(), path)?;
let file = OpenOptions::new().read(true).write(true).open(path)?;
fs::copy(source.path(), building_path(path))?;
let file = OpenOptions::new().read(true).write(true).open(building_path(path))?;
let mmap = unsafe { MmapMut::map_mut(&file)? };
let n = source.len();
Ok(Self {
@@ -441,7 +441,8 @@ impl PersistentBitVecBuilder {
}
pub fn close(self) -> io::Result<()> {
self.mmap.flush()
self.mmap.flush()?;
publish_built(&self.path)
}
pub fn finish(self) -> io::Result<PersistentBitVec> {
+3 -2
View File
@@ -15,7 +15,7 @@ use std::path::{Path, PathBuf};
use memmap2::{Mmap, MmapMut};
use crate::mmap_file::{create_mmap_file, open_mmap_file};
use crate::mmap_file::{create_mmap_file, open_mmap_file, publish_built};
const MAGIC: [u8; 4] = *b"PFIV";
@@ -178,7 +178,8 @@ impl PersistentFixedIntVecBuilder {
}
pub fn close(self) -> io::Result<()> {
self.mmap.flush()
self.mmap.flush()?;
publish_built(&self.path)
}
pub fn finish(self) -> io::Result<PersistentFixedIntVec> {
+1
View File
@@ -35,6 +35,7 @@ pub use intmatrix::{
};
pub use layer_meta::LayerMeta;
pub use mmap_file::{building_path, publish_built};
pub use matrix_builder::{ColBuilder, MatrixBuilder};
pub use rankselect::{PersistentRankSelectBitVec, PersistentRankSelectBitVecBuilder};
pub use reader::{Iter as CompactIntVecIter, PersistentCompactIntVec};
+26 -4
View File
@@ -10,13 +10,35 @@
//! `block_ranks`) and its own higher-level API — those genuinely differ
//! between formats and aren't forced into a shared shape.
use std::fs::{File, OpenOptions};
use std::fs::{self, File, OpenOptions};
use std::io::{self, Seek, SeekFrom, Write as _};
use std::path::Path;
use std::path::{Path, PathBuf};
use memmap2::{Mmap, MmapMut};
/// Creates (or truncates) `path`, writes the standard header — `magic` +
/// Suffix of the name a file lives under while it is being built.
const BUILDING_SUFFIX: &str = ".building";
/// The name `path` is built under: a file only appears at `path` itself once
/// complete ([`publish_built`]). A process killed mid-build leaves at most a
/// `.building` file, never a `path` with a valid header and unwritten data
/// that `path.exists()` would take for a finished one — what every
/// "skip it, it already exists" caller (per-layer annexes) relies on.
pub fn building_path(path: &Path) -> PathBuf {
let mut name = path.as_os_str().to_owned();
name.push(BUILDING_SUFFIX);
PathBuf::from(name)
}
/// Atomically moves the file built under [`building_path`]`(path)` to `path`
/// (same directory, so a plain `rename`), replacing any previous one. To be
/// called once the data is complete and flushed.
pub fn publish_built(path: &Path) -> io::Result<()> {
fs::rename(building_path(path), path)
}
/// Creates (or truncates) [`building_path`]`(path)` — the caller publishes it
/// with [`publish_built`] once complete — writes the standard header — `magic` +
/// `pad` (4 bytes right after `magic`; usually all-zero, but
/// `PersistentFixedIntVec` repurposes byte 4 to store its `width`) +
/// `n as u64`, followed by `extra_header_zeros` more zero bytes for a
@@ -37,7 +59,7 @@ pub(crate) fn create_mmap_file(
.write(true)
.create(true)
.truncate(true)
.open(path)?;
.open(building_path(path))?;
file.write_all(&magic)?;
file.write_all(&pad)?;
file.write_all(&(n as u64).to_le_bytes())?;
+3 -2
View File
@@ -21,7 +21,7 @@ use std::path::{Path, PathBuf};
use common_traits::SelectInWord;
use memmap2::{Mmap, MmapMut};
use crate::mmap_file::{create_mmap_file, open_mmap_file};
use crate::mmap_file::{create_mmap_file, open_mmap_file, publish_built};
const MAGIC: [u8; 4] = *b"PRSB";
@@ -268,7 +268,8 @@ impl PersistentRankSelectBitVecBuilder {
let block_off = HEADER_SIZE + bits_bytes(self.n);
let block_bytes = u64_slice_to_le_bytes(&block_ranks);
self.mmap[block_off..block_off + block_bytes.len()].copy_from_slice(&block_bytes);
self.mmap.flush()
self.mmap.flush()?;
publish_built(&self.path)
}
pub fn finish(self) -> io::Result<PersistentRankSelectBitVec> {
+23
View File
@@ -548,6 +548,29 @@ impl PersistentSparseCompactIntMatrixBuilder {
Ok(builder)
}
/// Row-permuted copy of `src`: row `s` of the result is row
/// `new_to_old[s]` of `src`. Same contract as
/// [`PersistentSparseBitMatrixBuilder::build_permuted`].
pub fn build_permuted(
src: &PersistentSparseCompactIntMatrix,
new_to_old: &[u32],
dir: &Path,
) -> io::Result<Self> {
let mut builder = Self::new(new_to_old.len(), src.n_cols(), dir)?;
let mut cols: Vec<u32> = Vec::new();
let mut vals: Vec<u32> = Vec::new();
for &old in new_to_old {
cols.clear();
vals.clear();
src.for_each_cell_in_row(old as usize, |c, v| {
cols.push(c as u32);
vals.push(v);
});
builder.push_row(&cols, &vals);
}
Ok(builder)
}
pub fn close(self) -> io::Result<()> {
let Self {
support,
+28
View File
@@ -415,3 +415,31 @@ fn partial_jaccard_and_hamming_match_dense_including_diagonal() {
drop(dir_dense);
drop(dir_sparse);
}
#[test]
fn build_permuted_follows_the_permutation() {
// Mix of singleton and multi rows, including a repeated multi set, so the
// dictionary dedup is exercised on the permuted order too.
let rows: Vec<&[bool]> = vec![
&[true, false, false, false],
&[false, false, true, false],
&[true, true, false, false],
&[false, false, true, true],
&[true, true, false, false],
];
let (_dir, src) = make_sparse(&rows, 4);
let new_to_old: Vec<u32> = vec![3, 0, 4, 1, 2];
let out = tempdir().unwrap();
let permuted =
PersistentSparseBitMatrixBuilder::build_permuted(&src, &new_to_old, &out.path().join("p"))
.unwrap()
.finish()
.unwrap();
assert_eq!(permuted.n(), src.n());
assert_eq!(permuted.n_cols(), src.n_cols());
for (new, &old) in new_to_old.iter().enumerate() {
assert_eq!(row_as_bool(&permuted, new), row_as_bool(&src, old as usize), "new row {new}");
}
}
@@ -304,3 +304,32 @@ fn pack_sparse_compact_int_matrix_from_already_packed_matrix() {
assert_eq!(&*sparse.row(slot), &**expected, "slot {slot}");
}
}
#[test]
fn build_permuted_follows_the_permutation() {
// Rows sharing a support with different values, plus singletons.
let rows: Vec<&[u32]> = vec![
&[7, 0, 0, 0],
&[0, 0, 3, 0],
&[2, 5, 0, 0],
&[9, 1, 0, 0],
&[0, 0, 4, 6],
];
let (_dir, src) = make_sparse(&rows, 4);
let new_to_old: Vec<u32> = vec![4, 2, 0, 3, 1];
let out = tempdir().unwrap();
let permuted = PersistentSparseCompactIntMatrixBuilder::build_permuted(
&src,
&new_to_old,
&out.path().join("p"),
)
.unwrap()
.finish()
.unwrap();
assert_eq!(permuted.n(), src.n());
for (new, &old) in new_to_old.iter().enumerate() {
assert_eq!(permuted.row(new), src.row(old as usize), "new row {new}");
}
}
-2
View File
@@ -6,11 +6,9 @@ edition = "2021"
[dependencies]
obikseq = { path = "../obikseq" }
obifastwrite = { path = "../obifastwrite" }
ahash = "0.8"
hashbrown = { version = "0.17", features = ["rayon"] }
rayon = "1"
crossbeam-channel = "0.5"
xxhash-rust = { version = "0.8.15", features = ["xxh3", "const_xxh3"] }
tracing = "0.1"
[features]
+70 -41
View File
@@ -1,19 +1,19 @@
use hashbrown::HashMap;
use hashbrown::{HashMap, HashSet};
use obikseq::k;
use obikseq::{CanonicalKmer, Unitig};
#[cfg(not(any(test, feature = "test-utils")))]
use rayon::iter::{IntoParallelRefIterator, ParallelIterator};
use std::cell::RefCell;
use std::sync::atomic::{AtomicU8, Ordering};
use xxhash_rust::xxh3::Xxh3Builder;
use super::kmer_hash::KmerHashBuilder;
use super::node::{IS_VISITED_MASK, Node};
use super::unitig_iter::UnitigNucIter;
use super::walk::WalkState;
// ── Types ─────────────────────────────────────────────────────────────────────
pub(super) type FastHashMap<K, V> = HashMap<K, V, Xxh3Builder>;
pub(super) type FastHashMap<K, V> = HashMap<K, V, KmerHashBuilder>;
// ── GraphDeBruijn ─────────────────────────────────────────────────────────────
@@ -24,7 +24,7 @@ pub struct GraphDeBruijn {
impl GraphDeBruijn {
pub fn new() -> Self {
Self {
nodes: FastHashMap::with_hasher(Xxh3Builder::new()),
nodes: FastHashMap::with_hasher(KmerHashBuilder),
}
}
@@ -40,7 +40,7 @@ impl GraphDeBruijn {
pub fn with_capacity(capacity: usize) -> Self {
Self {
nodes: FastHashMap::with_capacity_and_hasher(capacity, Xxh3Builder::new()),
nodes: FastHashMap::with_capacity_and_hasher(capacity, KmerHashBuilder),
}
}
@@ -79,7 +79,7 @@ impl GraphDeBruijn {
if node.is_visited() {
return;
}
if self.is_start(*kmer, node) {
if self.is_start(*kmer, node, atomic, true) || self.is_start(*kmer, node, atomic, false) {
node.set_start();
atomic.store(node.0, Ordering::Relaxed);
}
@@ -98,45 +98,40 @@ impl GraphDeBruijn {
}
}
pub fn start_walk(&self, kmer: CanonicalKmer) -> WalkState {
let node = Node(
self.nodes
.get(&kmer)
.expect("Cannot start a walk from not indexed kmer")
.load(Ordering::Relaxed),
);
pub fn start_walk(&self, kmer: CanonicalKmer) -> WalkState<'_> {
let atomic = self
.nodes
.get(&kmer)
.expect("Cannot start a walk from not indexed kmer");
let node = Node(atomic.load(Ordering::Relaxed));
debug_assert!(
!node.is_visited(),
"Cannot start a walk from a visited node"
);
WalkState::new(kmer, node, true)
WalkState::new(kmer, node, true, atomic)
}
pub fn try_start_walk(&self, kmer: CanonicalKmer) -> Option<WalkState> {
let node = Node(self.nodes.get(&kmer)?.load(Ordering::Relaxed));
pub fn try_start_walk(&self, kmer: CanonicalKmer) -> Option<WalkState<'_>> {
let atomic = self.nodes.get(&kmer)?;
let node = Node(atomic.load(Ordering::Relaxed));
if node.is_visited() {
return None;
}
Some(WalkState::new(kmer, node, true))
Some(WalkState::new(kmer, node, true, atomic))
}
pub(super) fn unitig_nucleotides(&self, kmer: CanonicalKmer, k: usize) -> Option<UnitigNucIter<'_>> {
let old = self
.nodes
.get(&kmer)?
.fetch_or(IS_VISITED_MASK, Ordering::AcqRel);
pub(super) fn unitig_nucleotides(&self, kmer: CanonicalKmer, k: usize, direct: bool) -> Option<UnitigNucIter<'_>> {
let atomic = self.nodes.get(&kmer)?;
let old = atomic.fetch_or(IS_VISITED_MASK, Ordering::AcqRel);
if old & IS_VISITED_MASK != 0 {
return None;
}
let start = WalkState::new(kmer, Node(old), true);
let start = WalkState::new(kmer, Node(old), direct, atomic);
let next_step = start.walk(self).and_then(|(next_state, nuc)| {
let ext_old = self
.nodes
.get(&next_state.kmer)?
.fetch_or(IS_VISITED_MASK, Ordering::AcqRel);
let ext_old = next_state.atomic.fetch_or(IS_VISITED_MASK, Ordering::AcqRel);
(ext_old & IS_VISITED_MASK == 0).then_some((next_state, nuc))
});
Some(UnitigNucIter::new(self, kmer, k, next_step))
Some(UnitigNucIter::new(self, kmer, k, direct, next_step))
}
pub fn for_each_unitig(&self, f: impl Fn(UnitigNucIter<'_>) + Sync) {
@@ -157,7 +152,8 @@ impl GraphDeBruijn {
.then_some(kmer)
})
.for_each(|kmer| {
if let Some(iter) = self.unitig_nucleotides(kmer, k) {
let Some(direct) = self.start_direction(kmer) else { return };
if let Some(iter) = self.unitig_nucleotides(kmer, k, direct) {
n_new.fetch_add(1, Ordering::Relaxed);
f(iter);
}
@@ -166,7 +162,8 @@ impl GraphDeBruijn {
#[cfg(any(test, feature = "test-utils"))]
self.nodes.iter().for_each(|(&kmer, atomic)| {
if Node(atomic.load(Ordering::Acquire)).is_start() {
if let Some(iter) = self.unitig_nucleotides(kmer, k) {
let Some(direct) = self.start_direction(kmer) else { return };
if let Some(iter) = self.unitig_nucleotides(kmer, k, direct) {
n_new.fetch_add(1, Ordering::Relaxed);
f(iter);
}
@@ -187,8 +184,8 @@ impl GraphDeBruijn {
if node.is_visited() {
continue;
}
let chain_start = self.find_chain_start(kmer);
if let Some(iter) = self.unitig_nucleotides(chain_start, k) {
let (chain_start, direct) = self.find_chain_start(kmer);
if let Some(iter) = self.unitig_nucleotides(chain_start, k, direct) {
n2.fetch_add(1, Ordering::Relaxed);
f(iter);
}
@@ -196,7 +193,7 @@ impl GraphDeBruijn {
// Safe because unitig_nucleotides may have visited kmer in the
// meantime — in that case it returns None and we skip.
if chain_start != kmer {
if let Some(iter) = self.unitig_nucleotides(kmer, k) {
if let Some(iter) = self.unitig_nucleotides(kmer, k, true) {
n2.fetch_add(1, Ordering::Relaxed);
f(iter);
}
@@ -216,17 +213,34 @@ impl GraphDeBruijn {
self.nodes.extend(other.nodes);
}
fn find_chain_start(&self, kmer: CanonicalKmer) -> CanonicalKmer {
let node = Node(self.nodes[&kmer].load(Ordering::Acquire));
let mut state = WalkState::new(kmer, node, false);
let mut seen = std::collections::HashSet::new();
/// Walks backwards from `kmer` to the first node of its chain. Returns
/// that node together with the orientation to read it in to walk the chain
/// forwards: the backward walk traverses nodes in the orientation
/// `state.direct`, so the forward one is its opposite. (Dropping the
/// orientation, as this function once did, started the forward walk at the
/// wrong end and emitted a lone-kmer unitig.) On a cycle, `kmer` itself
/// in the canonical orientation.
fn find_chain_start(&self, kmer: CanonicalKmer) -> (CanonicalKmer, bool) {
let atomic = &self.nodes[&kmer];
let node = Node(atomic.load(Ordering::Acquire));
let mut state = WalkState::new(kmer, node, false, atomic);
// Most nodes reaching this pass are branch nodes or tips that cannot
// step at all: answer before allocating the cycle-detection set.
let Some((first, _)) = state.walk(self) else {
return (state.kmer, !state.direct);
};
let mut seen: HashSet<(u64, bool), KmerHashBuilder> = HashSet::with_hasher(KmerHashBuilder);
seen.insert((state.kmer.raw(), state.direct));
if !seen.insert((first.kmer.raw(), first.direct)) {
return (kmer, true);
}
state = first;
loop {
match state.walk(self) {
None => return state.kmer,
None => return (state.kmer, !state.direct),
Some((next, _)) => {
if !seen.insert((next.kmer.raw(), next.direct)) {
return kmer;
return (kmer, true);
}
state = next;
}
@@ -234,8 +248,23 @@ impl GraphDeBruijn {
}
}
fn is_start(&self, query: CanonicalKmer, node: Node) -> bool {
!WalkState::new(query, node, true).reachable(self)
/// Orientation in which a node flagged as a start must be read: the
/// canonical one if a unitig starts there that way, else the reverse
/// complement. `None` if it was visited in the meantime.
fn start_direction(&self, kmer: CanonicalKmer) -> Option<bool> {
let atomic = self.nodes.get(&kmer)?;
let node = Node(atomic.load(Ordering::Acquire));
if node.is_visited() {
return None;
}
Some(self.is_start(kmer, node, atomic, true) || !self.is_start(kmer, node, atomic, false))
}
/// `true` when `query`, traversed in orientation `direct` (canonical or
/// reverse-complemented), has no unique predecessor — i.e. a unitig
/// starts there when read in that orientation.
fn is_start(&self, query: CanonicalKmer, node: Node, atomic: &AtomicU8, direct: bool) -> bool {
!WalkState::new(query, node, direct, atomic).reachable(self)
}
pub fn try_for_each_unitig<E, F>(&self, f: F) -> Result<(), E>
+55
View File
@@ -0,0 +1,55 @@
//! Hasher for the graph's `HashMap<CanonicalKmer, _>`.
//!
//! The key is one `u64` (the derived `Hash` of a kmer is a single
//! `write_u64`), so the generic streaming hashers used before (xxh3's
//! `Hasher` implementation: buffer, update, finish) cost far more than the
//! hash itself — ≈ 35 % of the active CPU of `index` in the profile of
//! `DevDocMD/implementation/mphf_batch_queries.md`, §17. A 64-bit finaliser
//! (splitmix64's) is a bijection with full avalanche, which is what
//! hashbrown needs from it (low bits for the bucket, top 7 for the tag).
use std::hash::{BuildHasher, Hasher};
#[derive(Clone, Copy, Default)]
pub(super) struct KmerHashBuilder;
impl BuildHasher for KmerHashBuilder {
type Hasher = KmerHasher;
#[inline]
fn build_hasher(&self) -> KmerHasher {
KmerHasher(0)
}
}
pub(super) struct KmerHasher(u64);
#[inline]
fn mix64(x: u64) -> u64 {
let x = x ^ (x >> 30);
let x = x.wrapping_mul(0xbf58476d1ce4e5b9);
let x = x ^ (x >> 27);
let x = x.wrapping_mul(0x94d049bb133111eb);
x ^ (x >> 31)
}
impl Hasher for KmerHasher {
#[inline]
fn write_u64(&mut self, x: u64) {
self.0 = mix64(self.0 ^ x);
}
/// Generic fallback (never taken for a kmer key): folds the bytes 8 at a time.
fn write(&mut self, bytes: &[u8]) {
for chunk in bytes.chunks(8) {
let mut buf = [0u8; 8];
buf[..chunk.len()].copy_from_slice(chunk);
self.write_u64(u64::from_le_bytes(buf));
}
}
#[inline]
fn finish(&self) -> u64 {
self.0
}
}
+2 -1
View File
@@ -1,10 +1,11 @@
//! De Bruijn graph over canonical k-mers, built for unitig extraction.
//!
//! Submodules: [`node`] (packed per-kmer neighbour/visited/start flags),
//! Submodules: [`kmer_hash`] (hasher of the node table), [`node`] (packed per-kmer neighbour/visited/start flags),
//! [`walk`] (single-step traversal), [`graph`] ([`GraphDeBruijn`] itself),
//! [`unitig_iter`] (nucleotide-by-nucleotide unitig walk iterator).
mod graph;
mod kmer_hash;
mod node;
mod unitig_iter;
mod walk;
+12 -8
View File
@@ -10,9 +10,11 @@ use super::walk::WalkState;
pub struct UnitigNucIter<'a> {
graph: &'a GraphDeBruijn,
start: CanonicalKmer,
/// Orientation the unitig is read in (`false`: reverse complement of `start`).
direct: bool,
pos: usize,
k: usize,
next_step: Option<(WalkState, u8)>,
next_step: Option<(WalkState<'a>, u8)>,
}
impl<'a> UnitigNucIter<'a> {
@@ -20,11 +22,13 @@ impl<'a> UnitigNucIter<'a> {
graph: &'a GraphDeBruijn,
start: CanonicalKmer,
k: usize,
next_step: Option<(WalkState, u8)>,
direct: bool,
next_step: Option<(WalkState<'a>, u8)>,
) -> Self {
Self {
graph,
start,
direct,
pos: 0,
k,
next_step,
@@ -37,16 +41,16 @@ impl Iterator for UnitigNucIter<'_> {
fn next(&mut self) -> Option<u8> {
if self.pos < self.k {
let nuc = self.start.nucleotide(self.pos);
let nuc = if self.direct {
self.start.nucleotide(self.pos)
} else {
3 - self.start.nucleotide(self.k - 1 - self.pos)
};
self.pos += 1;
Some(nuc)
} else if let Some((state, nuc)) = self.next_step.take() {
self.next_step = state.walk(self.graph).and_then(|(next_state, next_nuc)| {
let old = self
.graph
.nodes
.get(&next_state.kmer)?
.fetch_or(IS_VISITED_MASK, Ordering::AcqRel);
let old = next_state.atomic.fetch_or(IS_VISITED_MASK, Ordering::AcqRel);
(old & IS_VISITED_MASK == 0).then_some((next_state, next_nuc))
});
Some(nuc)
+22 -12
View File
@@ -1,35 +1,41 @@
use obikseq::{CanonicalKmer, Sequence};
use std::sync::atomic::Ordering;
use std::sync::atomic::{AtomicU8, Ordering};
use super::graph::GraphDeBruijn;
use super::node::Node;
pub struct WalkState {
/// A position in a walk: the node's kmer, its flags as read when the step was
/// taken, and the node's own table entry. Keeping `atomic` means the caller
/// that wants to mark the node visited does not look the kmer up a second
/// time (a random access in a table far larger than the cache).
pub struct WalkState<'g> {
pub(super) kmer: CanonicalKmer,
pub(super) node: Node,
pub(super) direct: bool,
pub(super) atomic: &'g AtomicU8,
}
impl WalkState {
pub fn new(kmer: CanonicalKmer, node: Node, direct: bool) -> Self {
impl<'g> WalkState<'g> {
pub fn new(kmer: CanonicalKmer, node: Node, direct: bool, atomic: &'g AtomicU8) -> Self {
debug_assert!(!node.is_visited(), "Cannot walk over a visited node");
Self { kmer, node, direct }
Self { kmer, node, direct, atomic }
}
pub fn leavable(&self, graph: &GraphDeBruijn) -> bool {
pub fn leavable(&self, graph: &'g GraphDeBruijn) -> bool {
self.walk(graph).is_some()
}
pub fn reachable(&self, graph: &GraphDeBruijn) -> bool {
pub fn reachable(&self, graph: &'g GraphDeBruijn) -> bool {
WalkState {
kmer: self.kmer,
node: self.node,
direct: !self.direct,
atomic: self.atomic,
}
.leavable(graph)
}
pub fn walk(&self, graph: &GraphDeBruijn) -> Option<(WalkState, u8)> {
pub fn walk(&self, graph: &'g GraphDeBruijn) -> Option<(WalkState<'g>, u8)> {
if self.direct {
if !self.node.can_extend_right() {
return None;
@@ -38,7 +44,8 @@ impl WalkState {
let next = self.kmer.into_kmer().push_right(nuc);
let cnext = next.canonical();
let dnext = next.raw() == cnext.raw();
let next_node = Node(graph.nodes.get(&cnext).unwrap().load(Ordering::Relaxed));
let next_atomic = graph.nodes.get(&cnext).unwrap();
let next_node = Node(next_atomic.load(Ordering::Relaxed));
if next_node.is_visited() {
return None;
}
@@ -52,6 +59,7 @@ impl WalkState {
kmer: cnext,
node: next_node,
direct: dnext,
atomic: next_atomic,
},
nuc,
))
@@ -63,20 +71,22 @@ impl WalkState {
let next = self.kmer.into_kmer().push_left(nuc);
let cnext = next.canonical();
let dnext = next.raw() != cnext.raw();
let next_node = Node(graph.nodes.get(&cnext).unwrap().load(Ordering::Relaxed));
let next_atomic = graph.nodes.get(&cnext).unwrap();
let next_node = Node(next_atomic.load(Ordering::Relaxed));
if next_node.is_visited() {
return None;
}
let reachable = if dnext {
next_node.can_extend_right()
} else {
next_node.can_extend_left()
} else {
next_node.can_extend_right()
};
reachable.then_some((
WalkState {
kmer: cnext,
node: next_node,
direct: dnext,
atomic: next_atomic,
},
3 - nuc,
))
+83
View File
@@ -317,3 +317,86 @@ fn branching_graph_no_kmer_lost_or_duplicated() {
);
assert_eq!(got, expected, "k-mer sets differ");
}
// ── orientation independence ──────────────────────────────────────────────
// Deterministic pseudo-random ASCII sequence (xorshift), so the tests need no
// RNG dependency.
fn pseudo_random_ascii(seed: u64, len: usize) -> Vec<u8> {
let mut st = seed.wrapping_mul(0x9E3779B97F4A7C15) | 1;
(0..len)
.map(|_| {
st ^= st << 13;
st ^= st >> 7;
st ^= st << 17;
b"ACGT"[(st % 4) as usize]
})
.collect()
}
#[test]
fn linear_chain_is_one_unitig_whatever_the_orientations() {
// A repeat-free sequence is one maximal unitig. Each k-mer is stored in
// its canonical orientation, so across many sequences the chain's nodes
// take every mix of orientations — including chains whose two ends are
// both stored reverse-complemented, which once came out as several
// unitigs (a lone-kmer unitig per mis-oriented chain start) depending
// only on the table's iteration order.
let k = 21; // long enough that (k-1)-mer collisions (= branches) do not occur
let _guard = lock_k();
set_k(k);
for seed in 1..=300u64 {
let seq = pseudo_random_ascii(seed, 90);
let kmers = canonical_kmers(&seq);
if kmers.len() != seq.len() - k + 1 {
continue; // accidental internal repeat: not a single chain
}
let g = graph_from_ascii(&seq);
g.compute_degrees_and_mark_starts();
let unitigs = collect_unitigs(&g);
assert_eq!(unitigs.len(), 1, "seed {seed}: repeat-free sequence must give one unitig");
assert_eq!(unitigs[0].seql(), seq.len(), "seed {seed}: unitig spans the sequence");
assert_eq!(kmers_from_unitigs(&unitigs), kmers, "seed {seed}: same k-mers");
}
}
#[test]
fn unitigs_partition_the_kmers_of_branching_reads() {
// Reads of a small genome on both strands with substitutions: branches,
// bubbles and tips. Whatever the shape, every k-mer must be covered
// exactly once: sum over unitigs of (seql - k + 1) == number of nodes.
let k = 11;
let _guard = lock_k();
set_k(k);
let genome = pseudo_random_ascii(7, 600);
let mut g = GraphDeBruijn::new();
for r in 0..120u64 {
let start = (pseudo_random_ascii(100 + r, 1)[0] as usize * 31 + r as usize * 17) % 500;
let mut read = genome[start..start + 90].to_vec();
if r % 2 == 1 {
// second strand
read = read
.iter()
.rev()
.map(|&b| match b {
b'A' => b'T',
b'C' => b'G',
b'G' => b'C',
_ => b'A',
})
.collect();
}
if r % 3 == 0 {
let at = (r as usize * 13) % read.len();
read[at] = if read[at] == b'A' { b'C' } else { b'A' };
}
for i in 0..=read.len() - k {
g.push(Kmer::from_ascii(&read[i..i + k]).unwrap().canonical());
}
}
g.compute_degrees_and_mark_starts();
let unitigs = collect_unitigs(&g);
let covered: usize = unitigs.iter().map(|u| u.seql() - k + 1).sum();
assert_eq!(covered, g.len(), "every k-mer in exactly one unitig");
assert_eq!(kmers_from_unitigs(&unitigs).len(), g.len(), "no k-mer missing");
}
-19
View File
@@ -188,11 +188,6 @@ impl IndexCache {
self.iter().flat_map(|l| l.iter_unitigs())
}
#[inline]
pub fn hash(&self, partition: usize, layer: usize, kmer: CanonicalKmer) -> Option<usize> {
Some(self.get_layer(partition, layer)?.hash(kmer))
}
pub fn n_kmer(&self, partition: usize, layer: usize) -> Option<usize> {
Some(self.get_layer(partition, layer)?.n())
}
@@ -228,20 +223,6 @@ impl IndexCache {
None
}
/// Membership-checked lookup — routes `kmer` straight to its one
/// possible partition (`CanonicalKmer::partition`, the same
/// minimizer-hash rule every k-mer was scattered by when the index was
/// built) instead of scanning every cached partition: a k-mer can only
/// ever live in the partition it hashes to, so anywhere else is not
/// worth even asking. Uses `self.index().n_partitions()` — the index's
/// real total, not [`n_partition`](Self::n_partition) (this cache's own
/// count, which can be a subset — see [`new`](Self::new)). `None` if
/// that partition isn't cached or doesn't carry `kmer`.
pub fn find(&self, kmer: CanonicalKmer) -> Option<(usize, usize, usize)> {
let partition = kmer.partition(self.raw_index.n_partitions());
self.find_in_partition(partition, kmer)
}
/// Parallel map-reduce over every cached `KmerLayer` (across every
/// cached partition), bounded to `max_concurrent` layers scanned at
/// once — a cap on how many layers are actively being mapped
+3 -3
View File
@@ -12,9 +12,9 @@ obicompactvec = { path = "../obicompactvec" }
obidebruinj = { path = "../obidebruinj" }
obipipeline = { path = "../obipipeline" }
obikentropy = { path = "../obikentropy" }
cacheline-ef = "1.1"
epserde = "0.8"
ptr_hash = "1.1"
cacheline-ef = "2.0"
epserde = "0.13"
ptr_hash = { version = "2.1", features = ["epserde"] }
niffler = "3.0.0"
memmap2 = "0.9.10"
ndarray = "0.17"
+4 -2
View File
@@ -90,7 +90,7 @@ impl IndexBuilder for KmerIndex {
let idx = KmerIndex {
root_path: output.to_owned(),
meta: Arc::new(meta),
lock: std::sync::Mutex::new(None),
lock: obisys::LockSlot::empty(output),
};
if !idx.acquire_lock() {
return Err(OKIError::Io(std::io::Error::other(
@@ -102,7 +102,9 @@ impl IndexBuilder for KmerIndex {
fn finalize_indexed<P: AsRef<Path>>(output: P, rep: &mut Reporter, sparse: bool) -> OKIResult<Self> {
let output = output.as_ref();
let idx = KmerIndex::open(output)?;
// `output` is held exclusively by the builder that wrote it: reopen
// without a lock, a locking `open` would wait for that builder for ever.
let idx = KmerIndex::open_unlocked(output)?;
idx.meta.mark_indexed().map_err(OKIError::Io)?;
let t_pack = Stage::start("pack");
idx.pack_matrices(sparse)?;
+86 -56
View File
@@ -1,6 +1,6 @@
use std::fs;
use std::path::{Path, PathBuf};
use std::sync::{Arc, Mutex};
use std::sync::Arc;
use crate::layer::KmerLayer;
use crate::partition::KmerPartition;
@@ -15,7 +15,7 @@ use crate::index::meta::{GenomeInfo, IndexConfig, IndexMeta};
pub struct KmerIndex {
pub(crate) root_path: PathBuf,
pub(crate) meta: Arc<IndexMeta>,
pub(crate) lock: Mutex<Option<obisys::DirLock>>,
pub(crate) lock: obisys::LockSlot,
}
impl KmerIndex {
@@ -37,11 +37,8 @@ impl KmerIndex {
set_m(config.minimizer_size);
let genomes = genome_info.into_iter().collect();
let meta = IndexMeta::create_at(&root_path, config, genomes).map_err(OKIError::Io)?;
let idx = Self {
root_path,
meta: Arc::new(meta),
lock: Mutex::new(None),
};
let lock = obisys::LockSlot::empty(&root_path);
let idx = Self { root_path, meta: Arc::new(meta), lock };
if !idx.acquire_lock() {
return Err(OKIError::Io(std::io::Error::other(
"failed to acquire index lock right after create",
@@ -50,67 +47,92 @@ impl KmerIndex {
Ok(idx)
}
/// Open an existing index for reading — doesn't touch the directory
/// lock at all: concurrent readers (`query`, `dump`, a `merge` source,
/// ...) must never block on each other. A caller that needs exclusive
/// access (resuming an interrupted build) uses
/// [`open_lock`](Self::open_lock) instead; either way, `self.lock`
/// starts empty — [`test_lock`](Self::test_lock)/
/// [`acquire_lock`](Self::acquire_lock) are opt-in from here.
/// Open an existing index for reading, holding a **shared** directory
/// lock for as long as the returned index lives: any number of readers
/// coexist, but a writer ([`open_lock`](Self::open_lock),
/// [`with_lock`](Self::with_lock)) waits for them all, and readers wait
/// for a writer at work — a reader never sees files change under it.
/// The lock is taken before the metadata is read.
///
/// Never open, from a process that holds the write lock on a directory
/// through another handle, that same directory again with `open`: the
/// new handle would wait for the first for ever.
pub fn open<P: AsRef<Path>>(path: P) -> OKIResult<Self> {
let root_path = path.as_ref().to_owned();
let meta = IndexMeta::open_at(&root_path).map_err(OKIError::Io)?;
set_k(meta.config.kmer_size);
set_m(meta.config.minimizer_size);
Ok(Self {
root_path,
meta: Arc::new(meta),
lock: Mutex::new(None),
})
Self::open_with(path.as_ref(), obisys::LockMode::Shared)
}
/// `open`, then hold the directory lock exclusively — for a caller
/// that resumes a write pipeline (e.g. `obikmer index` re-run on an
/// existing, not-yet-`Indexed` directory) rather than reading
/// concurrently with others.
/// `open`, holding the directory lock **exclusively** for as long as the
/// returned index lives — **the** way for a command that modifies an
/// existing index in place (`convert`, `pack`, `annotate`, `utils
/// rename`, resuming an interrupted build, ...) to claim it. The lock is
/// taken *before* the metadata is read, so what the caller reads is what
/// it will modify.
///
/// A command that only reads, or that writes a *new* index elsewhere
/// (`merge`, `filter`, `select`: exclusive on their destination, shared
/// on their sources), uses [`open`](Self::open). Library code that has to
/// write into an index it was handed (annexes built on demand) wraps that
/// write in [`with_lock`](Self::with_lock).
pub fn open_lock<P: AsRef<Path>>(path: P) -> OKIResult<Self> {
let idx = Self::open(path)?;
if !idx.acquire_lock() {
return Err(OKIError::Io(std::io::Error::other(
"failed to acquire index lock",
)));
Self::open_with(path.as_ref(), obisys::LockMode::Exclusive)
}
fn open_with(path: &Path, mode: obisys::LockMode) -> OKIResult<Self> {
// A directory that is not an index is reported by `open_unlocked`,
// without a lock file being created in it.
if !Self::is_an_index(path) {
return Self::open_unlocked(path);
}
let lock = match mode {
obisys::LockMode::Shared => obisys::LockSlot::shared(path),
obisys::LockMode::Exclusive => obisys::LockSlot::exclusive(path),
}
.map_err(OKIError::Io)?;
let mut idx = Self::open_unlocked(path)?;
idx.lock = lock;
Ok(idx)
}
/// Block until this index's directory lock is acquired, then hold it
/// (released on [`release_lock`](Self::release_lock) or when `self` is
/// dropped). Returns `false` only on a genuine I/O failure acquiring
/// the lock file itself — contention blocks rather than failing.
pub fn acquire_lock(&self) -> bool {
match obisys::DirLock::acquire(&self.root_path) {
Ok(lock) => {
*self.lock.lock().unwrap() = Some(lock);
true
}
Err(_) => false,
}
/// `open` without taking any lock — for a caller that already holds the
/// directory through another handle (`finalize_indexed`, reopening the
/// index its own builder holds exclusively).
pub(crate) fn open_unlocked(path: &Path) -> OKIResult<Self> {
let root_path = path.to_owned();
let meta = IndexMeta::open_at(&root_path).map_err(OKIError::Io)?;
set_k(meta.config.kmer_size);
set_m(meta.config.minimizer_size);
let lock = obisys::LockSlot::empty(&root_path);
Ok(Self { root_path, meta: Arc::new(meta), lock })
}
/// Take the exclusive directory lock and hold it until `self` is dropped
/// — for an index just created, which is about to be written to.
/// Idempotent. Returns `false` only on a genuine I/O failure.
pub(crate) fn acquire_lock(&self) -> bool {
self.lock.acquire_exclusive().is_ok()
}
/// Whether this index object currently holds its directory lock
/// exclusively (a shared lock is not "write access").
pub fn holds_write_lock(&self) -> bool {
self.lock.holds_write_lock()
}
/// Run `f` with the directory lock held **exclusively** — promoting the
/// shared lock a handle from [`open`](Self::open) holds, and restoring it
/// afterwards; see [`obisys::LockSlot::with_exclusive`], including what
/// `f` must re-check. Library code that writes into an index it was
/// handed (annexes built on demand) wraps that write in this.
pub fn with_lock<T>(&self, f: impl FnOnce() -> OKIResult<T>) -> OKIResult<T> {
self.lock.with_exclusive(f)
}
/// Non-blocking probe of this index's directory lock — doesn't retain
/// it either way: `true` if the lock was free (briefly acquired, then
/// immediately released), `false` if another process currently holds
/// it. For a read-only caller that wants to warn about a build in
/// progress without itself becoming exclusive; use
/// [`acquire_lock`](Self::acquire_lock)/[`open_lock`](Self::open_lock)
/// to actually hold it.
/// anything: `true` if no process currently holds it exclusively, `false`
/// if one does (this process included). For a caller that wants to warn
/// about a write in progress without waiting for it.
pub fn test_lock(&self) -> bool {
matches!(obisys::DirLock::try_acquire(&self.root_path), Ok(Some(_)))
}
/// Release this index's directory lock, if held. A no-op if it isn't.
pub fn release_lock(&self) {
*self.lock.lock().unwrap() = None;
self.lock.is_free_of_writer()
}
/// Return `true` if `path` points to a valid obikmer index directory.
@@ -210,6 +232,14 @@ impl KmerIndex {
crate::partition::index_dir(&self.root_path, i)
}
/// Path of layer `j` of partition `i` — pure path computation, nothing
/// is opened (unlike [`layer`](Self::layer)/`layer_dir`, which open the
/// layer's MPHF). For maintenance tools that must reach a layer's files
/// without going through a format they can't read yet.
pub fn layer_path(&self, i: usize, j: usize) -> PathBuf {
crate::layer::utils::layer_dir(&self.index_dir(i), j)
}
/// Number of layers in partition `i` — a disk scan, not a metadata read
/// (see [`KmerPartition::n_layers`](crate::partition::KmerPartition::n_layers)).
pub fn n_layers(&self, i: usize) -> OKIResult<usize> {
+4 -2
View File
@@ -280,8 +280,10 @@ impl IndexMeta {
// ── private ──────────────────────────────────────────────────────────────
fn read_full(root: &Path) -> io::Result<IndexMetadata> {
let file = fs::File::open(root.join(META_FILENAME))?;
serde_json::from_reader(file).map_err(io::Error::other)
// One `read`, then parse from memory: `from_reader` on a bare `File`
// issues one syscall per byte (≈ 0.35 s for a 91-genome `index.meta`).
let bytes = fs::read(root.join(META_FILENAME))?;
serde_json::from_slice(&bytes).map_err(io::Error::other)
}
fn write_full(&self, on_disk: &IndexMetadata) -> io::Result<()> {
+11 -9
View File
@@ -199,6 +199,17 @@ impl KmerLayer {
}
}
/// Batch form of [`find_slot`](Self::find_slot): `out[i]` is
/// `find_slot(kmers[i])`, MPHF lookups prefetched together. `out` is
/// cleared and reused.
pub fn find_slot_batch(&self, kmers: &[CanonicalKmer], out: &mut Vec<Option<usize>>) {
match self {
KmerLayer::Count { layer, .. } => layer.find_slot_batch(kmers, out),
KmerLayer::Presence { layer, .. } => layer.find_slot_batch(kmers, out),
KmerLayer::Empty { .. } => panic!("Layer::find_slot_batch() called on an Empty layer"),
}
}
/// Raw MPHF batch lookup: kmer → slot, no membership check — for
/// callers that already know every kmer is a member of *this* layer, so
/// the evidence check `find_slot`/`find` would perform is redundant.
@@ -286,15 +297,6 @@ impl KmerLayer {
}
}
/// Raw MPHF lookup: kmer → slot, no membership check.
pub fn hash(&self, kmer: CanonicalKmer) -> usize {
match self {
KmerLayer::Count { layer, .. } => layer.hash(kmer),
KmerLayer::Presence { layer, .. } => layer.hash(kmer),
KmerLayer::Empty { .. } => panic!("Layer::hash() called on an Empty layer"),
}
}
/// Iterate over all canonical kmers in the layer, in deterministic order.
pub fn iter_kmers(&self) -> crate::layer::mphf_layer::KmerIter {
match self {
+1 -1
View File
@@ -27,7 +27,7 @@ pub enum IndexMode {
/// [`detect`](Self::detect)).
Approx { b: u8, z: u8 },
/// Hybrid: both `fingerprint.bin` and `evidence.bin` + `unitigs.bin.idx`.
/// `find()` uses the fingerprint (O(1), approx); `find_strict()` uses exact evidence.
/// `find()` uses the fingerprint (O(1), approx); the exact evidence is kept alongside.
Hybrid { b: u8, z: u8 },
}
+4 -1
View File
@@ -3,11 +3,14 @@ pub mod evidence;
pub mod fingerprint;
pub mod meta;
pub(crate) mod mphf_layer;
pub(crate) mod slotted;
pub mod typed_layer;
pub(crate) mod utils;
pub use crate::index::error::{OKIError, OKIResult};
pub use content_layer::KmerLayer;
pub use meta::IndexMode;
pub use mphf_layer::{EvidenceKind, KmerBatchIter, KmerIter, MphfLayer, MphfOnly, UnitigIter};
pub use mphf_layer::{
EvidenceKind, KmerBatchIter, KmerIter, Mphf, MphfLayer, MphfOnly, UnitigIter, mphf_params,
};
pub use typed_layer::{HasLayerContent, HasStorageKind, Hit, LayerContent, LayerData, TypedLayer};
+86 -101
View File
@@ -8,24 +8,29 @@ use epserde::prelude::*;
use obicompactvec::LayerMeta;
use obikseq::CanonicalKmer;
use obiskio::{CanonicalKmerIter, UnitigFileReader, UnitigFileWriter, build_unitig_idx};
use ptr_hash::{PtrHash, PtrHashParams, bucket_fn::CubicEps, hash::Xx64};
use ptr_hash::{PtrHash, PtrHashParams, bucket_fn::CubicEps, hash::Xxh3Int};
use crate::index::error::{OKIError, OKIResult};
use crate::layer::evidence::{Evidence, EvidenceWriter};
use crate::layer::fingerprint::{FingerprintVec, FingerprintVecWriter};
use crate::layer::meta::IndexMode;
use crate::layer::slotted::SlottedKmers;
pub(crate) const MPHF_FILE: &str = "mphf.bin";
pub(crate) const UNITIGS_FILE: &str = "unitigs.bin";
pub(crate) const EVIDENCE_FILE: &str = "evidence.bin";
pub(crate) const FINGERPRINT_FILE: &str = "fingerprint.bin";
/// Owned MPHF — used only at build time (construction + store).
pub(crate) type Mphf = PtrHash<u64, CubicEps, CachelineEfVec<Vec<CachelineEf>>, Xx64, Vec<u8>>;
/// The layer MPHF: `Xxh3Int` hasher, `CubicEps` buckets, multi-part (`SINGLE_PART = false`),
/// minimal (`REMAP = true`). Built owned (construction + store); queried through
/// `MemCase<Mphf>::uncase()`, the ε-copy view into the memory-mapped file.
pub type Mphf = PtrHash<u64, CubicEps, CachelineEfVec<Vec<CachelineEf>>, Xxh3Int, Vec<u8>, false, true>;
/// Zero-copy MPHF for querying — ε-deserialized view into a memory-mapped file.
/// `MemCase` owns the mmap backing; `'static` is sound because MemCase pins the memory.
type MphfEps = PtrHash<u64, CubicEps, CachelineEfVec<&'static [CachelineEf]>, Xx64, &'static [u8]>;
/// Construction parameters of [`Mphf`] — the one place they are chosen
/// (`lambda = 3.5`, `alpha = 0.99`, `CubicEps`, ~2.4 bits/key).
pub fn mphf_params() -> PtrHashParams<CubicEps> {
PtrHashParams::<CubicEps>::default_balanced()
}
// ── LayerEvidence ─────────────────────────────────────────────────────────────
@@ -46,17 +51,17 @@ enum LayerEvidence {
Hybrid {
evidence_path: PathBuf,
fingerprint_path: PathBuf,
unitig_path: PathBuf,
},
}
/// Resources actually opened (mmap). `Exact`/`Hybrid` carry their own
/// `unitigs` — needed to complete evidence verification, not shared with
/// `MphfLayer`'s own separate `unitigs` (used by `iter_kmers`/`Approx`'s
/// `find_strict` fallback, mode-independent). Yes, this means a layer's
/// `unitigs.bin` gets mmapped twice for Exact/Hybrid — deliberately not
/// optimised away here: avoiding that redundant open is the future cache
/// crate's job, not this base type's.
/// Resources actually opened (mmap). `Exact` carries its own `unitigs` —
/// needed to complete evidence verification, not shared with `MphfLayer`'s
/// own separate `unitigs` (used by `iter_kmers`, mode-independent). Yes, this
/// means an Exact layer's `unitigs.bin` gets mmapped twice (the `.idx`
/// offsets are mapped too, so both readers share the same pages) —
/// deliberately not optimised away here. `Hybrid` keeps its `evidence.bin`
/// open (it gives the slot count) but nothing verifies against it: lookups
/// use the fingerprint, as in `Approx`.
enum OpenLayerEvidence {
Exact {
evidence: Evidence,
@@ -68,7 +73,6 @@ enum OpenLayerEvidence {
Hybrid {
evidence: Evidence,
fingerprint: FingerprintVec,
unitigs: Arc<UnitigFileReader>,
},
}
@@ -88,7 +92,6 @@ impl LayerEvidence {
EvidenceKind::Hybrid => LayerEvidence::Hybrid {
evidence_path: layer_dir.join(EVIDENCE_FILE),
fingerprint_path: layer_dir.join(FINGERPRINT_FILE),
unitig_path: layer_dir.join(UNITIGS_FILE),
},
})
}
@@ -103,10 +106,9 @@ impl LayerEvidence {
LayerEvidence::Approx { fingerprint_path } => OpenLayerEvidence::Approx {
fingerprint: FingerprintVec::open(&fingerprint_path)?,
},
LayerEvidence::Hybrid { evidence_path, fingerprint_path, unitig_path } => OpenLayerEvidence::Hybrid {
LayerEvidence::Hybrid { evidence_path, fingerprint_path } => OpenLayerEvidence::Hybrid {
evidence: Evidence::open(&evidence_path)?,
fingerprint: FingerprintVec::open(&fingerprint_path)?,
unitigs: Arc::new(UnitigFileReader::open(&unitig_path)?),
},
})
}
@@ -146,19 +148,15 @@ impl EvidenceKind {
/// Autonomous kmer → slot mapping for one layer.
///
/// Two query methods:
/// - [`find`](Self::find) — O(1), uses fingerprint (Approx/Hybrid) or exact evidence (Exact).
/// - [`find_strict`](Self::find_strict) — always exact; O(1) on Exact/Hybrid layers,
/// O(n) sequential scan on Approx layers.
/// Query: [`find`](Self::find) — O(1), uses the fingerprint (Approx/Hybrid) or
/// exact evidence (Exact).
pub struct MphfLayer {
/// Mode-independent copy — used by `iter_kmers` (all modes) and
/// `find_strict`'s `Approx` fallback. `Exact`/`Hybrid` additionally
/// carry their own inside `OpenLayerEvidence` (see its doc comment for
/// why that's a second, deliberately unshared mmap of the same file,
/// not a bug).
/// Mode-independent copy — used by `iter_kmers` (all modes). `Exact`
/// additionally carries its own inside `OpenLayerEvidence` (see its doc
/// comment for why that's a second mmap of the same file, not a bug).
unitigs: Arc<UnitigFileReader>,
mphf: MemCase<MphfEps>,
mphf: MphfOnly,
ev: OpenLayerEvidence,
n: usize,
}
@@ -171,8 +169,7 @@ impl MphfLayer {
/// not stored: `KmerLayer`, one level up, is what owns a layer's
/// directory as a persistent field, not this type.
pub fn open(dir: &Path) -> OKIResult<Self> {
let mphf: MemCase<MphfEps> = Mphf::mmap(&dir.join(MPHF_FILE), Flags::empty())
.map_err(|e| OKIError::InvalidData { context: "layer", detail: e.to_string() })?;
let mphf = MphfOnly::open(dir)?;
let unitigs = Arc::new(UnitigFileReader::open(&dir.join(UNITIGS_FILE))?);
let ev = LayerEvidence::at(dir)?.open()?;
let n = match &ev {
@@ -188,10 +185,17 @@ impl MphfLayer {
/// O(1) lookup — dispatches automatically:
/// - Exact: evidence + `verify_canonical_kmer`, zero false positives.
/// - Approx: fingerprint check, false-positive rate ≈ 1/2^b.
/// - Hybrid: fingerprint check (fast path), zero false positives via `find_strict`.
/// - Hybrid: fingerprint check, same as Approx (the exact evidence is kept on disk, not consulted).
#[inline]
pub fn find(&self, kmer: CanonicalKmer) -> Option<usize> {
let slot = self.mphf.index(&kmer.raw());
self.confirm(self.mphf.index(kmer), kmer)
}
/// Membership check of `kmer` against the slot the MPHF gave it — the
/// mode-dependent half of [`find`](Self::find), shared with
/// [`find_batch`](Self::find_batch).
#[inline]
fn confirm(&self, slot: usize, kmer: CanonicalKmer) -> Option<usize> {
if slot >= self.n {
return None;
}
@@ -215,41 +219,20 @@ impl MphfLayer {
}
}
/// Always-exact lookup — zero false positives regardless of mode.
/// Batch form of [`find`](Self::find): `out[i]` is `find(kmers[i])`.
///
/// - Exact/Hybrid: O(1) via evidence + `verify_canonical_kmer`.
/// - Approx: O(n) sequential scan of `unitigs.bin` to confirm the kmer
/// that owns the slot, then exact comparison.
pub fn find_strict(&self, kmer: CanonicalKmer) -> Option<usize> {
let slot = self.mphf.index(&kmer.raw());
if slot >= self.n {
return None;
}
match &self.ev {
OpenLayerEvidence::Exact { evidence, unitigs }
| OpenLayerEvidence::Hybrid { evidence, unitigs, .. } => {
let (chunk_id, rank) = evidence.decode(slot);
if unitigs.verify_canonical_kmer(chunk_id as usize, rank as usize, kmer) {
Some(slot)
} else {
None
}
}
OpenLayerEvidence::Approx { .. } => {
// `self.unitigs` was opened via `UnitigFileReader::open`, which
// falls back to sequential mode automatically when no `.idx`
// sidecar exists — true for every Approx layer (`build_unitig_idx`
// is only ever called for Exact/Hybrid). Reusing it here avoids
// the second file-open + mmap + kmer-count scan a fresh
// `open_sequential` call would repeat on every call.
for (stored, _, _) in self.unitigs.iter_indexed_canonical_kmers() {
if self.mphf.index(&stored.raw()) == slot {
return if stored == kmer { Some(slot) } else { None };
}
}
None
}
}
/// The MPHF lookups of the whole slice go through `ptr_hash`'s prefetching
/// stream (see [`MphfOnly::for_each_slot`]), so their random pilot reads
/// overlap instead of being paid one after the other; the membership
/// check runs on each slot as it comes out. `out` is cleared and reused.
pub fn find_batch(&self, kmers: &[CanonicalKmer], out: &mut Vec<Option<usize>>) {
out.clear();
out.reserve(kmers.len());
let mut i = 0;
self.mphf.for_each_slot(kmers, |slot| {
out.push(self.confirm(slot, kmers[i]));
i += 1;
});
}
/// Number of slots in the MPHF layer — equal to the number of stored k-mers.
@@ -267,23 +250,13 @@ impl MphfLayer {
}
}
/// Raw MPHF lookup: kmer → slot.
///
/// Returns the slot assigned by the MPHF without performing any membership
/// check against the layer's evidence. The returned slot is only meaningful
/// when the kmer is actually present in the layer; callers that need an
/// existence test should use [`find`](Self::find) instead.
pub fn hash(&self, kmer: CanonicalKmer) -> usize {
self.mphf.index(&kmer.raw())
}
/// Batch raw MPHF lookup: kmers → slots.
///
/// Returns a [`Vec`] of slots, one per input kmer, in the same order as the
/// input slice. Like [`index`](Self::index), no membership check is
/// performed.
/// input slice. Like [`hash`](Self::hash), no membership check is
/// performed. Prefetching — see [`MphfOnly::index_batch`].
pub fn hash_batch(&self, kmers: &[CanonicalKmer]) -> Vec<usize> {
kmers.iter().map(|k| self.mphf.index(&k.raw())).collect()
self.mphf.index_batch(kmers)
}
/// Iterate over all canonical kmers in the layer, in deterministic order.
@@ -426,11 +399,11 @@ impl Iterator for KmerBatchIter {
///
/// Use this when the caller guarantees that all queried kmers are in the MPHF
/// domain (e.g. when iterating the source's own unitigs during merge).
pub struct MphfOnly(MemCase<MphfEps>);
pub struct MphfOnly(MemCase<Mphf>);
impl MphfOnly {
pub fn open(dir: &Path) -> OKIResult<Self> {
let mphf: MemCase<MphfEps> = Mphf::mmap(&dir.join(MPHF_FILE), Flags::empty())
let mphf: MemCase<Mphf> = unsafe { Mphf::mmap(&dir.join(MPHF_FILE), Flags::empty()) }
.map_err(|e| OKIError::InvalidData { context: "layer", detail: e.to_string() })?;
Ok(Self(mphf))
}
@@ -438,13 +411,30 @@ impl MphfOnly {
/// Return the slot for `kmer`. Only valid when `kmer` is in the MPHF domain.
#[inline]
pub fn index(&self, kmer: CanonicalKmer) -> usize {
self.0.index(&kmer.raw())
self.0.uncase().index(&kmer.raw())
}
/// Batch form of [`index`](Self::index) — one MPHF lookup pass over the
/// whole slice instead of one call per kmer.
/// Batch form of [`index`](Self::index): one prefetching `ptr_hash`
/// stream over the whole slice, so the random pilot reads of different
/// kmers overlap instead of being paid one after the other.
pub fn index_batch(&self, kmers: &[CanonicalKmer]) -> Vec<usize> {
kmers.iter().map(|&k| self.index(k)).collect()
let mut slots = Vec::with_capacity(kmers.len());
self.for_each_slot(kmers, |slot| slots.push(slot));
slots
}
/// Calls `f(slot)` for each kmer of `kmers`, in order.
///
/// Callback-shaped on purpose: `ptr_hash`'s `index_stream` only
/// implements `fold` (its `next()` is `unimplemented!`), so it cannot be
/// returned as an ordinary iterator without a panic trap for callers.
/// Look-ahead of 32 keys, as recommended by `ptr_hash`.
#[inline]
pub fn for_each_slot(&self, kmers: &[CanonicalKmer], f: impl FnMut(usize)) {
self.0
.uncase()
.index_stream::<32, _>(kmers.iter().map(|k| k.raw()))
.for_each(f);
}
}
@@ -468,14 +458,13 @@ impl MphfLayer {
return Ok(0);
}
let mphf: Mphf = Mphf::load_full(&dir.join(MPHF_FILE))
let mphf: Mphf = unsafe { Mphf::load_full(&dir.join(MPHF_FILE)) }
.map_err(|e| OKIError::InvalidData { context: "layer", detail: e.to_string() })?;
let mut ev = EvidenceWriter::new(n);
let mut seen = vec![0u8; (n + 7) / 8];
for (kmer, chunk_id, rank) in unitigs.iter_indexed_canonical_kmers() {
let slot = mphf.index(&kmer.raw());
for (slot, _, chunk_id, rank) in SlottedKmers::new(&mphf, unitigs.iter_indexed_canonical_kmers()) {
if slot >= n {
return Err(OKIError::InvalidData { context: "mphf", detail: "slot out of bounds".into() });
}
@@ -512,13 +501,12 @@ impl MphfLayer {
return Ok(0);
}
let mphf: Mphf = Mphf::load_full(&dir.join(MPHF_FILE))
let mphf: Mphf = unsafe { Mphf::load_full(&dir.join(MPHF_FILE)) }
.map_err(|e| OKIError::InvalidData { context: "layer", detail: e.to_string() })?;
let mut fw = FingerprintVecWriter::new(n, b);
for (kmer, _, _) in unitigs.iter_indexed_canonical_kmers() {
let slot = mphf.index(&kmer.raw());
for (slot, kmer, _, _) in SlottedKmers::new(&mphf, unitigs.iter_indexed_canonical_kmers()) {
if slot >= n {
return Err(OKIError::InvalidData { context: "mphf", detail: "slot out of bounds".into() });
}
@@ -552,9 +540,9 @@ impl MphfLayer {
// ── Empty layer ───────────────────────────────────────────────────────
if n == 0 {
let mphf: Mphf = Mphf::try_new(&[] as &[u64], PtrHashParams::<CubicEps>::default())
let mphf: Mphf = Mphf::try_new(&[] as &[u64], mphf_params())
.ok_or_else(|| OKIError::InvalidData { context: "mphf", detail: "construction failed".into() })?;
mphf.store(&dir.join(MPHF_FILE))
unsafe { mphf.store(&dir.join(MPHF_FILE)) }
.map_err(|e| OKIError::InvalidData { context: "layer", detail: e.to_string() })?;
match mode {
IndexMode::Exact | IndexMode::Hybrid { .. } => {
@@ -576,9 +564,9 @@ impl MphfLayer {
let mphf: Mphf = Mphf::new_from_par_iter(
n,
keys.map(|k| k.raw()).par_bridge(),
PtrHashParams::<CubicEps>::default(),
mphf_params(),
);
mphf.store(&dir.join(MPHF_FILE))
unsafe { mphf.store(&dir.join(MPHF_FILE)) }
.map_err(|e| OKIError::InvalidData { context: "layer", detail: e.to_string() })?;
// ── Pass 2: fill evidence files + callback ────────────────────────────
@@ -588,8 +576,7 @@ impl MphfLayer {
match mode {
IndexMode::Exact => {
let mut ev = EvidenceWriter::new(n);
for (kmer, chunk_id, rank) in unitigs2.iter_indexed_canonical_kmers() {
let slot = mphf.index(&kmer.raw());
for (slot, kmer, chunk_id, rank) in SlottedKmers::new(&mphf, unitigs2.iter_indexed_canonical_kmers()) {
if slot >= n {
return Err(OKIError::InvalidData { context: "mphf", detail: "slot out of bounds".into() });
}
@@ -608,8 +595,7 @@ impl MphfLayer {
IndexMode::Approx { b, .. } => {
let mut fw = FingerprintVecWriter::new(n, *b);
for (kmer, _, _) in unitigs2.iter_indexed_canonical_kmers() {
let slot = mphf.index(&kmer.raw());
for (slot, kmer, _, _) in SlottedKmers::new(&mphf, unitigs2.iter_indexed_canonical_kmers()) {
if slot >= n {
return Err(OKIError::InvalidData { context: "mphf", detail: "slot out of bounds".into() });
}
@@ -628,8 +614,7 @@ impl MphfLayer {
IndexMode::Hybrid { b, .. } => {
let mut ev = EvidenceWriter::new(n);
let mut fw = FingerprintVecWriter::new(n, *b);
for (kmer, chunk_id, rank) in unitigs2.iter_indexed_canonical_kmers() {
let slot = mphf.index(&kmer.raw());
for (slot, kmer, chunk_id, rank) in SlottedKmers::new(&mphf, unitigs2.iter_indexed_canonical_kmers()) {
if slot >= n {
return Err(OKIError::InvalidData { context: "mphf", detail: "slot out of bounds".into() });
}
+58
View File
@@ -0,0 +1,58 @@
//! A layer's kmers streamed together with their MPHF slot — the shape every
//! build pass needs (`slot` indexes evidence, fingerprint and bitmap writes).
use obikseq::CanonicalKmer;
use crate::layer::mphf_layer::Mphf;
/// Kmers per `ptr_hash` stream: large enough to amortise its ring-buffer
/// start-up, small enough to stay in L1/L2 and never hold the layer.
const CHUNK: usize = 4096;
/// `(slot, kmer, chunk_id, rank)` for every `(kmer, chunk_id, rank)` of the
/// wrapped iterator, in the same order. The MPHF lookups are done a chunk at
/// a time through `ptr_hash`'s prefetching stream, so their random pilot
/// reads overlap; memory stays bounded by one chunk.
pub(crate) struct SlottedKmers<'a, I> {
mphf: &'a Mphf,
kmers: I,
chunk: Vec<(CanonicalKmer, usize, usize)>,
slots: Vec<usize>,
pos: usize,
}
impl<'a, I: Iterator<Item = (CanonicalKmer, usize, usize)>> SlottedKmers<'a, I> {
pub(crate) fn new(mphf: &'a Mphf, kmers: I) -> Self {
Self {
mphf,
kmers,
chunk: Vec::with_capacity(CHUNK),
slots: Vec::with_capacity(CHUNK),
pos: 0,
}
}
}
impl<I: Iterator<Item = (CanonicalKmer, usize, usize)>> Iterator for SlottedKmers<'_, I> {
type Item = (usize, CanonicalKmer, usize, usize);
fn next(&mut self) -> Option<Self::Item> {
if self.pos == self.chunk.len() {
self.chunk.clear();
self.chunk.extend(self.kmers.by_ref().take(CHUNK));
if self.chunk.is_empty() {
return None;
}
self.slots.clear();
let slots = &mut self.slots;
self.mphf
.index_stream::<32, _>(self.chunk.iter().map(|(k, _, _)| k.raw()))
.for_each(|slot| slots.push(slot));
self.pos = 0;
}
let (kmer, chunk_id, rank) = self.chunk[self.pos];
let slot = self.slots[self.pos];
self.pos += 1;
Some((slot, kmer, chunk_id, rank))
}
}
+7 -5
View File
@@ -189,13 +189,15 @@ impl<D: LayerData> TypedLayer<D> {
self.kmerstore.find(kmer)
}
pub fn n(&self) -> usize {
self.kmerstore.n()
/// Batch form of [`find_slot`](Self::find_slot): `out[i]` is
/// `find_slot(kmers[i])`, with the MPHF lookups prefetched together
/// (see `MphfLayer::find_batch`). `out` is cleared and reused.
pub fn find_slot_batch(&self, kmers: &[CanonicalKmer], out: &mut Vec<Option<usize>>) {
self.kmerstore.find_batch(kmers, out)
}
/// Raw MPHF lookup: kmer → slot, no membership check.
pub fn hash(&self, kmer: CanonicalKmer) -> usize {
self.kmerstore.hash(kmer)
pub fn n(&self) -> usize {
self.kmerstore.n()
}
/// Batch raw MPHF lookup: kmers → slots, no membership check.
+1 -3
View File
@@ -14,9 +14,7 @@ obicompactvec = { path = "../obicompactvec" }
obipipeline = { path = "../obipipeline" }
obiread = { path = "../obiread" }
obiskbuilder = { path = "../obiskbuilder" }
cacheline-ef = "1.1"
epserde = "0.8"
ptr_hash = "1.1"
epserde = "0.13"
niffler = "3.0.0"
memmap2 = "0.9.10"
rayon = "1"
+165
View File
@@ -0,0 +1,165 @@
//! Throwaway benchmark of the de Bruijn graph on a real kmer set:
//! `graph_bench <unitigs.bin> <k> [repeat]` builds `GraphDeBruijn` from the
//! canonical kmers of an existing layer's unitigs (same call sequence as
//! `write_graph_as_unitigs`) and times each phase.
use std::path::PathBuf;
use std::time::Instant;
use obidebruinj::GraphDeBruijn;
use obikseq::Sequence;
use obiskio::CanonicalKmerIter;
use rayon::prelude::*;
use std::collections::HashSet;
fn main() {
let args: Vec<String> = std::env::args().collect();
let path = PathBuf::from(&args[1]);
let k: usize = args[2].parse().unwrap();
let repeat: usize = args.get(3).map_or(1, |s| s.parse().unwrap());
obikseq::set_k(k);
obikseq::set_m(11);
let kmers: Vec<_> = if let Ok(n) = std::env::var("SYNTH") {
// synthetic reads: random genome, reads on both strands, 1 % substitutions
let n: usize = n.parse().unwrap();
let mut st = 0x9E3779B97F4A7C15u64;
let mut rnd = move || {
st ^= st << 13;
st ^= st >> 7;
st ^= st << 17;
st
};
let genome: Vec<u8> = (0..n).map(|_| b"ACGT"[(rnd() % 4) as usize]).collect();
let mut set = std::collections::BTreeSet::new();
for _ in 0..(n / 50) {
let start = (rnd() as usize) % (n - 100);
let mut read: Vec<u8> = genome[start..start + 100].to_vec();
for b in read.iter_mut() {
if rnd() % 100 == 0 {
*b = b"ACGT"[(rnd() % 4) as usize];
}
}
for w in read.windows(k) {
set.insert(obikseq::Kmer::from_ascii(w).unwrap().canonical());
}
}
set.into_iter().collect()
} else {
CanonicalKmerIter::new(&path).unwrap().collect()
};
eprintln!("{} kmers", kmers.len());
// Independent count of maximal unitigs: nodes − mutually-unique links
// (cycles ignored), computed with a plain HashSet — no use of GraphDeBruijn.
if std::env::var("IDEAL").is_ok() {
let set: HashSet<u64> = kmers.iter().map(|c| c.raw()).collect();
let outs = |a: obikseq::Kmer| -> Vec<obikseq::Kmer> {
(0..4u8)
.map(|x| a.push_right(x))
.filter(|b| set.contains(&b.canonical().raw()))
.collect()
};
let links: usize = kmers
.par_iter()
.map(|c| {
let mut n = 0usize;
for o in [c.into_kmer(), c.into_kmer().revcomp()] {
let ob = outs(o);
if ob.len() == 1 {
let b = ob[0];
// in-neighbours of b = revcomp of out-neighbours of revcomp(b)
if outs(b.revcomp()).len() == 1 {
// count each undirected link once: only from the orientation
// whose canonical raw is the smaller end (ties: both seen once per orientation)
let (ca, cb) = (o.canonical().raw(), b.canonical().raw());
if ca < cb || (ca == cb && o.raw() <= b.raw()) {
n += 1;
} else if ca > cb {
// seen from the other end too: skip
}
}
}
}
n
})
.sum();
eprintln!("IDEAL: nodes={} links={} -> maximal unitigs ≈ {}", kmers.len(), links, kmers.len() - links);
}
if std::env::var("IDEALDUMP").is_ok() {
let set: HashSet<u64> = kmers.iter().map(|c| c.raw()).collect();
let outs = |a: obikseq::Kmer| -> Vec<obikseq::Kmer> {
(0..4u8).map(|x| a.push_right(x)).filter(|b| set.contains(&b.canonical().raw())).collect()
};
// oriented kmer a has a mutual predecessor p iff ins(a)=={p} and outs(p)=={a}
let pred = |a: obikseq::Kmer| -> Option<obikseq::Kmer> {
let ins: Vec<obikseq::Kmer> = outs(a.revcomp()).into_iter().map(|b| b.revcomp()).collect();
if ins.len() == 1 && outs(ins[0]).len() == 1 { Some(ins[0]) } else { None }
};
let mut seen_chains = std::collections::BTreeSet::new();
for c in &kmers {
for o in [c.into_kmer(), c.into_kmer().revcomp()] {
if pred(o).is_some() { continue; }
let mut text: Vec<u8> = (0..k).map(|i| b"ACGT"[o.nucleotide(i) as usize]).collect();
let mut cur = o;
loop {
let ob = outs(cur);
if ob.len() != 1 { break; }
let b = ob[0];
if pred(b).is_none() { break; }
text.push(b"ACGT"[b.nucleotide(k - 1) as usize]);
cur = b;
}
let s = String::from_utf8(text).unwrap();
let rc: String = s.bytes().rev().map(|c| match c { b'A' => 'T', b'C' => 'G', b'G' => 'C', _ => 'A' }).collect();
seen_chains.insert(if s <= rc { s } else { rc });
}
}
for u in &seen_chains { println!("IDEALUNITIG {} {}", u.len(), u); }
}
for r in 0..repeat {
let t = Instant::now();
let mut g = GraphDeBruijn::new();
for &kmer in &kmers {
g.push(kmer);
}
let t_push = t.elapsed();
let t = Instant::now();
g.compute_degrees_and_mark_starts();
let t_deg = t.elapsed();
let t = Instant::now();
let n_unitigs = std::sync::atomic::AtomicUsize::new(0);
let n_nuc = std::sync::atomic::AtomicUsize::new(0);
let dump = std::env::var("DUMP").is_ok();
let seqs = std::sync::Mutex::new(Vec::<String>::new());
g.for_each_unitig(|it| {
let mut n = 0usize;
let mut text = String::new();
for nuc in it {
n += 1;
if dump {
text.push(b"ACGT"[nuc as usize] as char);
}
}
if dump {
seqs.lock().unwrap().push(text);
}
n_unitigs.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
n_nuc.fetch_add(n, std::sync::atomic::Ordering::Relaxed);
});
let t_walk = t.elapsed();
if dump {
let mut v = seqs.lock().unwrap().clone();
v.sort();
for u in v { println!("UNITIG {} {}", u.len(), u); }
}
println!(
"run {r}: nodes={} push={:.2?} degrees={:.2?} unitigs={:.2?} -> {} unitigs, {} nucleotides",
g.len(),
t_push,
t_deg,
t_walk,
n_unitigs.load(std::sync::atomic::Ordering::Relaxed),
n_nuc.load(std::sync::atomic::Ordering::Relaxed)
);
}
}
@@ -3,18 +3,15 @@ use std::fs;
use std::io;
use std::path::Path;
use cacheline_ef::{CachelineEf, CachelineEfVec};
use epserde::ser::Serialize as EpSerialize;
use memmap2::Mmap;
use obicompactvec::PersistentCompactIntVecBuilder;
use obikindex::layer::{Mphf, mphf_params};
use obiskio::{SKFileReader, SKResult};
use ptr_hash::{PtrHash, PtrHashParams, bucket_fn::CubicEps, hash::Xx64};
use tracing::debug;
use super::kmer_sort::sort_unique_kmers;
pub(super) type Mphf = PtrHash<u64, CubicEps, CachelineEfVec<Vec<CachelineEf>>, Xx64, Vec<u8>>;
fn build_mphf(unique_path: &Path, f0: usize) -> io::Result<Mphf> {
let file = fs::File::open(unique_path)?;
let mmap = unsafe { Mmap::map(&file)? };
@@ -24,7 +21,7 @@ fn build_mphf(unique_path: &Path, f0: usize) -> io::Result<Mphf> {
// Sequential constructor: the outer par_iter over partitions already saturates
// the Rayon pool. new_from_par_iter would get no additional threads and adds
// coordination overhead. try_new accesses the same mmap'd pages at zero extra cost.
Mphf::try_new(kmers, PtrHashParams::<CubicEps>::default())
Mphf::try_new(kmers, mphf_params())
.ok_or_else(|| io::Error::other("ptr_hash construction failed"))
}
@@ -73,7 +70,7 @@ pub(super) fn count_partition(dir: &Path, dedup_path: &Path, chunk_kmers: usize)
)
.map_err(io::Error::other)?;
EpSerialize::store(&mphf, &dir.join("mphf1.bin"))
unsafe { EpSerialize::store(&mphf, &dir.join("mphf1.bin")) }
.map_err(|e| io::Error::other(e.to_string()))?;
Ok(())
@@ -25,17 +25,13 @@ use std::io;
use std::path::Path;
use crate::graph_pipeline::{materialize_layer, write_graph_as_unitigs};
use cacheline_ef::{CachelineEf, CachelineEfVec};
use epserde::prelude::*;
use obicompactvec::{PersistentIntMatrix, PersistentCompactIntVec};
use obidebruinj::GraphDeBruijn;
use obikindex::layer::IndexMode;
use obikindex::layer::{KmerLayer, TypedLayer};
use obikindex::layer::{KmerLayer, Mphf, TypedLayer};
use obikindex::{KmerIndex, OKIError, OKIResult};
use obiskio::{SKError, SKFileMeta, SKFileReader};
use ptr_hash::{PtrHash, bucket_fn::CubicEps, hash::Xx64};
type Mphf = PtrHash<u64, CubicEps, CachelineEfVec<Vec<CachelineEf>>, Xx64, Vec<u8>>;
pub(crate) trait PrivateBuilder {
/// Mark scatter as complete (`IndexState::Scattered`).
@@ -157,7 +153,7 @@ impl PrivateBuilder for KmerIndex {
let mphf1_opt: Option<Mphf> = if need_counts {
let p = layer0_dir.join("mphf1.bin");
p.exists().then(|| Mphf::load_full(&p).ok()).flatten()
p.exists().then(|| unsafe { Mphf::load_full(&p) }.ok()).flatten()
} else {
None
};
+1 -1
View File
@@ -1,6 +1,6 @@
[package]
name = "obikmer"
version = "1.3.6"
version = "1.4.1"
edition = "2024"
[[bin]]
+5 -1
View File
@@ -93,7 +93,11 @@ fn run_annotate(args: &AnnotateArgs) {
}
};
let idx = open_index(&args.index);
// Rewrites the index metadata in place: exclusive for the whole command.
let idx = KmerIndex::open_lock(&args.index).unwrap_or_else(|e| {
eprintln!("error opening index: {e}");
std::process::exit(1);
});
let mut genomes = idx.meta().genomes().unwrap_or_else(|e| {
eprintln!("error reading index metadata: {e}");
+2 -8
View File
@@ -70,14 +70,8 @@ struct ModeArgs {
}
pub fn run(args: ConvertArgs) {
// Modifies the index in place; acquired before opening so a concurrent
// writer can't slip in between the open and the convert below.
let _lock = obisys::DirLock::acquire(&args.index).unwrap_or_else(|e| {
eprintln!("error locking index directory {}: {e}", args.index.display());
std::process::exit(1);
});
let idx = KmerIndex::open(&args.index).unwrap_or_else(|e| {
// Modifies the index in place: exclusive for the whole command.
let idx = KmerIndex::open_lock(&args.index).unwrap_or_else(|e| {
eprintln!("error opening index: {e}");
std::process::exit(1);
});
+2 -8
View File
@@ -26,14 +26,8 @@ pub struct PackArgs {
}
pub fn run(args: PackArgs) {
// Modifies the index in place; acquired before opening so a concurrent
// writer can't slip in between the open and the pack below.
let _lock = obisys::DirLock::acquire(&args.index).unwrap_or_else(|e| {
eprintln!("error locking index directory {}: {e}", args.index.display());
std::process::exit(1);
});
let idx = Arc::new(KmerIndex::open(&args.index).unwrap_or_else(|e| {
// Modifies the index in place: exclusive for the whole command.
let idx = Arc::new(KmerIndex::open_lock(&args.index).unwrap_or_else(|e| {
eprintln!("error opening index: {e}");
std::process::exit(1);
}));
+78 -3
View File
@@ -101,6 +101,51 @@ pub enum DistanceArg {
SnpTv,
}
/// The distance matrix this invocation must compute, if any: the one asked for
/// with `--distance`, else `jaccard` when `--nj`, `--upgma` or
/// `--shared-kmers` needs a matrix to work from, else none — a matrix is never
/// computed (nor written) just because nothing said otherwise.
pub fn distance_to_compute(
distance: Option<DistanceArg>,
nj: bool,
upgma: bool,
shared_kmers: bool,
) -> Option<DistanceArg> {
distance.or_else(|| (nj || upgma || shared_kmers).then_some(DistanceArg::Jaccard))
}
#[cfg(test)]
mod distance_tests {
use super::*;
#[test]
fn nothing_asked_nothing_computed() {
assert!(distance_to_compute(None, false, false, false).is_none());
}
#[test]
fn tree_and_shared_kmer_outputs_imply_jaccard() {
for (nj, upgma, shared) in [(true, false, false), (false, true, false), (false, false, true)] {
assert!(matches!(
distance_to_compute(None, nj, upgma, shared),
Some(DistanceArg::Jaccard)
));
}
}
#[test]
fn an_explicit_distance_wins() {
assert!(matches!(
distance_to_compute(Some(DistanceArg::Mash), true, true, true),
Some(DistanceArg::Mash)
));
assert!(matches!(
distance_to_compute(Some(DistanceArg::SnpJc), false, false, false),
Some(DistanceArg::SnpJc)
));
}
}
impl DistanceArg {
/// `Some` for the whole-index metrics, `None` for `snp-*` values.
pub fn as_classic(self) -> Option<DistanceMetric> {
@@ -215,6 +260,15 @@ pub struct PhyloArgs {
#[arg(long)]
pub shannon: bool,
/// Write the same entropy scan as `--shannon` as a distribution
/// (`<prefix>_shannon_summary.csv`): the number of families per entropy
/// kind, number of genomes carrying the family, number of those without
/// ambiguity and 0.05-bit entropy bin — a few hundred thousand lines at
/// most, whatever the size of the index, instead of one line per family.
/// Requires an already-built sibling annex, like `--shannon`.
#[arg(long)]
pub shannon_summary: bool,
/// Write a SNP-only pseudo-alignment (FASTA) — requires
/// `--n-sites <N|all>` and an already-built sibling annex
/// (`--sibling-annex` first, in this invocation or an earlier one).
@@ -260,6 +314,24 @@ pub struct PhyloArgs {
#[arg(long)]
pub pair_balance: bool,
/// Write what the draw contains, from the very sites it kept
/// (requires `--pseudo-alignment`, `--sankoff` or a `snp-*`
/// `--distance`): `<prefix>_sample_genomes.csv` (per genome, the sites it
/// has a call in and the informative sites it takes part in),
/// `<prefix>_sample_carriers.csv` (sites and informative sites per number
/// of carriers) and `<prefix>_sample_overlap.csv` (the sample's own
/// genome × genome overlap matrix, to compare with `--family-overlap`).
#[arg(long)]
pub sample_stats: bool,
/// Write how the draw was spread over the layers of the index
/// (requires `--pseudo-alignment`, `--sankoff` or a `snp-*`
/// `--distance`): `<prefix>_sample_plan.csv`, one line per layer with the
/// families eligible, their total weight, the layer's quota, the families
/// drawn and the ones kept after masking.
#[arg(long)]
pub sample_plan: bool,
/// Entropy-biased sampling target (Gaussian kernel mean) for
/// `--pseudo-alignment` — activates biasing as soon as this or
/// `--entropy-sd` is given; the other defaults to 1.0/0.5.
@@ -371,7 +443,10 @@ pub struct PhyloArgs {
#[arg(long, default_value = "100")]
pub sankoff_cost_scale: f64,
/// Distance to compute — either a whole-index metric (`jaccard`,
/// Distance matrix to compute and write — none is computed unless this
/// is given, or `--nj`/`--upgma`/`--shared-kmers` is (they need a matrix
/// and use `jaccard` then, see [`distance_to_compute`]). Either a
/// whole-index metric (`jaccard`,
/// `mash`, `hamming`, `bray-curtis`, ...) or a `snp-*` correction over
/// the central-position SNP substitution spectrum (`snp-raw`, `snp-jc`,
/// `snp-k2p`, `snp-k81`, `snp-f81`, `snp-t92`, `snp-tn93`, `snp-tv`) —
@@ -381,8 +456,8 @@ pub struct PhyloArgs {
/// `DevDocMD/theory/evolutionary_distances.md`, "`--distance`
/// unification" for the full catalog and why LogDet/Tajima-Nei/F84/
/// HKY85 aren't offered yet).
#[arg(long, value_enum, default_value = "jaccard")]
pub distance: DistanceArg,
#[arg(long, value_enum)]
pub distance: Option<DistanceArg>,
/// Rate-heterogeneity correction (Jin-Nei gamma shape parameter `α`)
/// for `snp-*` `--distance` values that support it
+60 -15
View File
@@ -3,6 +3,7 @@ mod iqtree;
mod phyg;
mod phylip;
mod sankoff;
mod stats;
mod tnt;
use std::io::{self, BufWriter, Write};
@@ -18,9 +19,10 @@ use obikphylo::{Metrics, neighbor_joining, upgma};
use obisys::{Reporter, Stage};
use tracing::info;
use args::NSitesArg;
use args::{DistanceArg, NSitesArg, distance_to_compute};
use iqtree::write_iqtree;
use phyg::write_sankoff_phyg;
use stats::{write_sample_plan, write_sample_stats};
use phylip::write_phylip_relaxed;
use sankoff::{write_sankoff_alignment_fasta, write_sankoff_matrix_csv, write_sankoff_params};
use tnt::write_sankoff_tnt;
@@ -76,13 +78,9 @@ pub fn run(args: PhyloArgs) {
// Meant to be (re)built routinely, ahead of any SNP-family distance
// computation that will later consume it.
if args.sibling_annex {
// Writes into the index directory — hold an exclusive lock for the
// duration so a second, concurrent `--sibling-annex` run on the
// same index can't corrupt these writes (see obisys::DirLock).
let _lock = obisys::DirLock::acquire(&args.index).unwrap_or_else(|e| {
eprintln!("error locking index directory {}: {e}", args.index.display());
std::process::exit(1);
});
// Writes into the index directory; `build_sibling_annex` holds the
// index's exclusive lock for the duration (`KmerIndex::with_lock`),
// so a concurrent run on the same index can't corrupt these writes.
info!("building sibling-count/minorant annex");
let t = Stage::start("sibling_annex");
cache.build_sibling_annex().unwrap_or_else(|e| {
@@ -194,6 +192,21 @@ pub fn run(args: PhyloArgs) {
info!("entropy report → {path}");
}
// ── Shannon entropy distribution (`--shannon-summary`) ──────────────────────
if args.shannon_summary {
let path = args.output.as_ref()
.map(|p| format!("{}_shannon_summary.csv", p.display()))
.unwrap_or_else(|| "shannon_summary.csv".into());
info!("summarising per-family Shannon entropy");
let t = Stage::start("shannon_summary");
cache.shannon_entropy_summary(std::path::Path::new(&path)).unwrap_or_else(|e| {
eprintln!("error summarising Shannon entropy: {e}");
std::process::exit(1);
});
rep.push(t.stop());
info!("entropy summary → {path}");
}
// ── `--min-shared-family` auto-exclusion ────────────────────────────────────
// Layered on top of `--exclude-genome`, not instead of it — a separate
// mask (not folded into `exclude_mask` itself) since it must reach
@@ -270,7 +283,8 @@ pub fn run(args: PhyloArgs) {
// different sample per flag — see `obikphylo::siblings::LayerReducer`'s
// own module docs).
let sankoff_requested = args.sankoff || args.tnt || args.phyg || args.iqtree;
let snp_kind = args.distance.as_snp();
let distance = distance_to_compute(args.distance, args.nj, args.upgma, args.shared_kmers);
let snp_kind = distance.and_then(DistanceArg::as_snp);
let need_alignment = args.pseudo_alignment || sankoff_requested;
let need_pairwise = sankoff_requested || snp_kind.is_some();
// `None` = every site (`--n-sites all`, or `--n-sites` omitted where it is optional)
@@ -284,10 +298,20 @@ pub fn run(args: PhyloArgs) {
eprintln!("error: --sankoff requires --n-sites <N|all>");
std::process::exit(1);
}
if (args.sample_stats || args.sample_plan) && !(need_alignment || need_pairwise) {
eprintln!(
"error: --sample-stats/--sample-plan describe a draw: they need --pseudo-alignment, \
--sankoff or a snp-* --distance"
);
std::process::exit(1);
}
let mut alignment_reducer =
need_alignment.then(|| obikphylo::siblings::AlignmentReducer::new(&cache, &snp_exclude_mask));
let mut pairwise_reducer = need_pairwise.then(|| obikphylo::siblings::PairwiseReducer::new(n));
let mut stats_reducer =
args.sample_stats.then(|| obikphylo::siblings::SampleStatsReducer::new(n, &snp_exclude_mask));
let mut sample_plan = None;
if need_alignment || need_pairwise {
let mut reducers: Vec<&mut dyn obikphylo::siblings::LayerReducer> = Vec::new();
@@ -297,6 +321,9 @@ pub fn run(args: PhyloArgs) {
if let Some(r) = pairwise_reducer.as_mut() {
reducers.push(r);
}
if let Some(r) = stats_reducer.as_mut() {
reducers.push(r);
}
info!(
"sampling SNP sites{}",
@@ -306,14 +333,24 @@ pub fn run(args: PhyloArgs) {
}
);
let t = Stage::start("sampled_scan");
obikphylo::siblings::sampled_scan(&cache, target, sampling, &snp_exclude_mask, &mut reducers)
.unwrap_or_else(|e| {
eprintln!("error sampling SNP sites: {e}");
std::process::exit(1);
});
sample_plan = Some(
obikphylo::siblings::sampled_scan(&cache, target, sampling, &snp_exclude_mask, &mut reducers)
.unwrap_or_else(|e| {
eprintln!("error sampling SNP sites: {e}");
std::process::exit(1);
}),
);
rep.push(t.stop());
}
// ── What the draw contains and how it was spread (`--sample-stats`/`--sample-plan`) ──
if let Some(reducer) = stats_reducer {
write_sample_stats(&reducer.finish(), &labels, &snp_exclude_mask, &args.output);
}
if args.sample_plan {
write_sample_plan(&sample_plan.expect("scanned above: need_alignment || need_pairwise"), &args.output);
}
// Finished once here (not left as reducers) so `--pseudo-alignment`
// and the Sankoff group can each read it below without fighting over
// which one consumes it — see `SankoffBundle::finish`'s own docs.
@@ -396,9 +433,17 @@ pub fn run(args: PhyloArgs) {
}
// ── Distance computation: classic whole-index metric vs. `snp-*` ───────────
// Nothing below (matrix, shared-kmer matrix, NJ/UPGMA trees) is wanted
// unless a distance was asked for (`--distance`) or implied
// (`--nj`/`--upgma`/`--shared-kmers`): stop here rather than compute and
// write a matrix nobody requested.
let Some(distance) = distance else {
rep.print();
return;
};
// Two genuinely different code paths behind one `--distance` value — see
// `args::DistanceArg`'s own docs.
let (matrix, shared_kmers) = match args.distance.as_classic() {
let (matrix, shared_kmers) = match distance.as_classic() {
Some(metric) => {
info!("computing {metric:?} distances for {n} genome(s)");
let need_shared = args.shared_kmers || args.nj || args.upgma;
+116
View File
@@ -0,0 +1,116 @@
// Diagnostic CSV outputs of a draw — `--sample-stats`, `--sample-plan` — kept
// out of `mod.rs` (already long) the way `tnt.rs`/`phyg.rs` hold their own
// writers. Each writer only formats what `obikphylo` computed; none of them
// decides anything.
use std::io::{BufWriter, Write};
use std::path::PathBuf;
use obikphylo::siblings::{SamplePlan, SampleStats};
use tracing::info;
/// `<prefix>_<suffix>`, or `<suffix>` alone without `-o` — the convention of
/// every other file `phylo` writes.
fn output_path(output: &Option<PathBuf>, suffix: &str) -> String {
match output {
Some(prefix) => format!("{}_{suffix}", prefix.display()),
None => suffix.to_string(),
}
}
fn create(path: &str) -> BufWriter<std::fs::File> {
BufWriter::new(std::fs::File::create(path).unwrap_or_else(|e| {
eprintln!("error creating {path}: {e}");
std::process::exit(1);
}))
}
/// `--sample-stats`: three CSV files describing what the draw contains.
///
/// - `_sample_genomes.csv`: per genome, the sampled sites it has a call in,
/// their share of all sampled sites, and the informative sites in which it
/// takes part in the bipartition; `excluded` genomes are listed with zeros.
/// - `_sample_carriers.csv`: per number of carriers, the sampled sites and how
/// many of them are informative.
/// - `_sample_overlap.csv`: the sample's own genome × genome overlap matrix
/// (sampled sites both genomes have a call in), same layout as
/// `--family-overlap`, to set against the index's.
pub fn write_sample_stats(
stats: &SampleStats,
labels: &[String],
excluded: &[bool],
output: &Option<PathBuf>,
) {
let n = stats.n_genomes;
let path = output_path(output, "sample_genomes.csv");
let mut f = create(&path);
writeln!(f, "genome,excluded,sites_with_call,share_of_sites,informative_sites").unwrap();
for g in 0..n {
let share = if stats.n_sites == 0 { 0.0 } else { stats.calls[g] as f64 / stats.n_sites as f64 };
writeln!(
f,
"{},{},{},{:.6},{}",
labels[g],
excluded.get(g).copied().unwrap_or(false),
stats.calls[g],
share,
stats.informative_calls[g]
)
.unwrap();
}
info!("sample genomes → {path}");
let path = output_path(output, "sample_carriers.csv");
let mut f = create(&path);
writeln!(f, "n_carriers,sites,informative_sites").unwrap();
for (carriers, (&sites, &informative)) in
stats.sites_by_carriers.iter().zip(&stats.informative_by_carriers).enumerate()
{
if sites > 0 {
writeln!(f, "{carriers},{sites},{informative}").unwrap();
}
}
info!("sample carriers → {path}");
let path = output_path(output, "sample_overlap.csv");
let mut f = create(&path);
write!(f, "genome").unwrap();
for label in labels.iter().take(n) {
write!(f, ",{label}").unwrap();
}
writeln!(f).unwrap();
for i in 0..n {
write!(f, "{}", labels[i]).unwrap();
for j in 0..n {
write!(f, ",{}", stats.overlap(i, j)).unwrap();
}
writeln!(f).unwrap();
}
info!("sample overlap matrix → {path}");
}
/// `--sample-plan`: `_sample_plan.csv`, one line per layer in the order the
/// draw visited them — the families eligible, their total weight, the layer's
/// quota, how many families were drawn and resolved, and how many survived
/// masking. Only layer lines (a column sum is one `awk` away); the target and
/// the total kept are logged.
pub fn write_sample_plan(plan: &SamplePlan, output: &Option<PathBuf>) {
let path = output_path(output, "sample_plan.csv");
let mut f = create(&path);
writeln!(f, "partition,layer,eligible,mass,quota,resolved,kept").unwrap();
for layer in &plan.layers {
writeln!(
f,
"{},{},{},{:.6},{},{},{}",
layer.partition, layer.layer, layer.eligible, layer.mass, layer.quota, layer.resolved, layer.kept
)
.unwrap();
}
info!(
"sample plan ({} layer(s), target {}, kept {}) → {path}",
plan.layers.len(),
plan.target,
plan.kept()
);
}
+24 -1
View File
@@ -192,17 +192,40 @@ pub fn run(args: QueryArgs) {
let overhead_multiplier =
BASE_OVERHEAD + n_genomes as u64 * BYTES_PER_KMER_PER_GENOME * detail_factor;
// Parallelism: a chunk is one worker's unit of work, so an input smaller
// than `n_workers` chunks leaves cores idle (two 48 MB files at the
// memory-derived 256 MiB cap = 2 chunks = 2 busy cores; measured 11.6 s
// against 2.5 s with 8 MiB chunks, see `DevDocMD/architecture/query.md`).
// When the input size can be estimated, aim for `CHUNKS_PER_WORKER` chunks
// per worker and never go below `MIN_CHUNK`. Stdin and URLs have no
// known size: the memory-derived size stands.
const CHUNKS_PER_WORKER: u64 = 2;
const MIN_CHUNK: u64 = 4 * 1024 * 1024;
let input_bytes: Option<u64> = args
.inputs
.iter()
.map(|p| obiread::estimated_decompressed_len(p))
.sum::<Option<u64>>();
let chunk_bytes = args
.chunk_size
.map(|mb| mb * 1024 * 1024)
.unwrap_or_else(|| {
let avail = available_memory_bytes();
let computed = avail / (n_workers as u64 * overhead_multiplier * SAFETY_FACTOR);
computed.clamp(4 * 1024 * 1024, 256 * 1024 * 1024) as usize
let by_memory = computed.clamp(4 * 1024 * 1024, 256 * 1024 * 1024);
match input_bytes {
Some(total) => {
let by_parallelism = (total / (n_workers as u64 * CHUNKS_PER_WORKER)).max(MIN_CHUNK);
by_memory.min(by_parallelism) as usize
}
None => by_memory as usize,
}
});
debug!(
chunk_bytes,
input_bytes,
n_genomes,
detail = args.detail,
overhead_multiplier,
+3 -1
View File
@@ -53,7 +53,9 @@ pub(super) fn run_bits_per_kmer(index_path: &PathBuf) {
pub(super) fn run_rename(index_path: &PathBuf, spec: &str) {
let (old_label, new_label) = parse_rename_spec(spec);
let idx = KmerIndex::open(index_path).unwrap_or_else(|e| {
// Rewrites the index metadata and renames a spectrum file in place:
// exclusive for the whole command.
let idx = KmerIndex::open_lock(index_path).unwrap_or_else(|e| {
eprintln!("error opening index: {e}");
std::process::exit(1);
});
-595
View File
@@ -1,595 +0,0 @@
//! Merging a source partition's new layer into a destination partition:
//! de Bruijn graph union (pass 1) then column fill (pass 2).
//!
//! Submodules: [`src_layer`] (`SrcLayerData`, the opened-source-matrix
//! lookup used by pass 2 here and by `rebuild_layer`). The `merge_partition`
//! orchestration itself stays in this file — its ~400-line body is one
//! tightly threaded pipeline (shared `Arc`/`Mutex` state across pass 1,
//! builder setup, and pass 2), not a set of independently callable steps.
use std::fs;
use std::io;
use std::path::{Path, PathBuf};
use std::sync::{Arc, Mutex};
use obipipeline::{
Pipeline, PipelineError, PipelineSender, SharedFlatFn, Stage, ThrottleGuard, WorkerPool,
make_sink, make_source, make_transform, throttle,
};
use tracing::debug;
use obicompactvec::{PersistentBitMatrixBuilder, PersistentCompactIntMatrixBuilder};
use obikseq::CanonicalKmer;
use obikindex::layer::{IndexMode, TypedLayer, LayeredMap, MphfOnly};
use obikindex::layer::utils::layer_dir;
use obiskio::UnitigFileReader;
use obikindex::{OKIError, OKIResult};
use obikindex::{ColBuilder, load_meta};
use obikindexer::{build_graph, materialize_layer};
use obikindex::KmerIndex;
mod src_layer;
pub use src_layer::SrcLayerData;
// ── MergeMode ─────────────────────────────────────────────────────────────────
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum MergeMode {
Presence,
Count,
}
// ── MatrixBuilder ─────────────────────────────────────────────────────────────
//
// Wraps whichever matrix builder `mode` calls for, so the merge pipeline never
// has to know the on-disk column naming (`col_NNNNNN.pbiv`/`.pciv`) or the
// matrix `meta.json` schema itself — both stay private to obicompactvec.
// `resume` reopens a matrix directory already closed by a previous builder
// session (an existing destination layer), continuing from its current
// `n_cols` instead of starting a fresh matrix at 0.
enum MatrixBuilder {
Bit(PersistentBitMatrixBuilder),
Int(PersistentCompactIntMatrixBuilder),
}
impl MatrixBuilder {
fn new(mode: MergeMode, n: usize, dir: &Path) -> io::Result<Self> {
Ok(match mode {
MergeMode::Presence => MatrixBuilder::Bit(PersistentBitMatrixBuilder::new(n, dir)?),
MergeMode::Count => MatrixBuilder::Int(PersistentCompactIntMatrixBuilder::new(n, dir)?),
})
}
fn resume(mode: MergeMode, dir: &Path) -> io::Result<Self> {
Ok(match mode {
MergeMode::Presence => MatrixBuilder::Bit(PersistentBitMatrixBuilder::resume(dir)?),
MergeMode::Count => MatrixBuilder::Int(PersistentCompactIntMatrixBuilder::resume(dir)?),
})
}
/// Add a column with no data written (all-zero/false) — for genome
/// columns absent from this source (e.g. dst genomes in a new layer).
fn add_absent_col(&mut self) -> io::Result<()> {
match self {
MatrixBuilder::Bit(b) => b.add_col()?.close(),
MatrixBuilder::Int(b) => b.add_col()?.close(),
}
}
fn add_col(&mut self) -> io::Result<ColBuilder> {
Ok(match self {
MatrixBuilder::Bit(b) => ColBuilder::Bit(b.add_col()?),
MatrixBuilder::Int(b) => ColBuilder::Int(b.add_col()?),
})
}
fn close(self) -> io::Result<()> {
match self {
MatrixBuilder::Bit(b) => b.close(),
MatrixBuilder::Int(b) => b.close(),
}
}
}
#[cfg(test)]
mod matrix_builder_tests {
use tempfile::tempdir;
use obicompactvec::{PersistentBitMatrix, PersistentCompactIntMatrix};
use super::{ColBuilder, MatrixBuilder, MergeMode};
/// Mirrors `merge_partition`'s "new layer" setup: absent (dst-genome)
/// columns get no data, then source columns are filled — all through one
/// continuous `MatrixBuilder` session, closed once at the end.
#[test]
fn new_layer_absent_then_source_columns_presence() {
let dir = tempdir().unwrap();
let data_dir = dir.path().join("presence");
let mut mb = MatrixBuilder::new(MergeMode::Presence, 3, &data_dir).unwrap();
// Two absent (dst-genome) columns.
mb.add_absent_col().unwrap();
mb.add_absent_col().unwrap();
// One source column, filled like pass 2 would.
let mut col = mb.add_col().unwrap();
match &mut col {
ColBuilder::Bit(b) => {
b.set(0, true);
b.set(1, false);
b.set(2, true);
}
ColBuilder::Int(_) => unreachable!(),
}
col.close().unwrap();
mb.close().unwrap();
let m = PersistentBitMatrix::open(dir.path()).unwrap();
assert_eq!(m.n_cols(), 3);
assert_eq!(&*m.row(0), &[false, false, true]);
assert_eq!(&*m.row(1), &[false, false, false]);
assert_eq!(&*m.row(2), &[false, false, true]);
}
#[test]
fn new_layer_absent_then_source_columns_count() {
let dir = tempdir().unwrap();
let data_dir = dir.path().join("counts");
let mut mb = MatrixBuilder::new(MergeMode::Count, 2, &data_dir).unwrap();
mb.add_absent_col().unwrap();
let mut col = mb.add_col().unwrap();
match &mut col {
ColBuilder::Int(b) => {
b.set(0, 7);
b.set(1, 42);
}
ColBuilder::Bit(_) => unreachable!(),
}
col.close().unwrap();
mb.close().unwrap();
let m = PersistentCompactIntMatrix::open(dir.path()).unwrap();
assert_eq!(m.n_cols(), 2);
assert_eq!(&*m.row(0), &[0u32, 7]);
assert_eq!(&*m.row(1), &[0u32, 42]);
}
/// Mirrors `merge_partition`'s "existing dst layer" setup: an
/// already-closed matrix (from a previous merge) gets more columns
/// appended via `resume`, without callers ever seeing `col_path`/
/// `MatrixMeta` — `n` itself is read back from the matrix directory.
#[test]
fn resume_appends_source_columns_to_existing_layer() {
let dir = tempdir().unwrap();
let data_dir = dir.path().join("presence");
// Previous merge: one dst-genome column already on disk.
let mut mb0 = MatrixBuilder::new(MergeMode::Presence, 3, &data_dir).unwrap();
mb0.add_absent_col().unwrap();
mb0.close().unwrap();
// This merge: resume and append two more source columns.
let mut mb = MatrixBuilder::resume(MergeMode::Presence, &data_dir).unwrap();
for vals in [[true, false, true], [false, true, false]] {
let mut col = mb.add_col().unwrap();
match &mut col {
ColBuilder::Bit(b) => {
for (slot, v) in vals.into_iter().enumerate() {
b.set(slot, v);
}
}
ColBuilder::Int(_) => unreachable!(),
}
col.close().unwrap();
}
mb.close().unwrap();
let m = PersistentBitMatrix::open(dir.path()).unwrap();
assert_eq!(m.n_cols(), 3);
assert_eq!(&*m.row(0), &[false, true, false]);
assert_eq!(&*m.row(1), &[false, false, true]);
assert_eq!(&*m.row(2), &[false, true, false]);
}
}
// ── KmerPartition::merge_partition ────────────────────────────────────────────
impl KmerIndex {
/// Merge `sources` into destination partition `i`.
///
/// Each entry in `sources` is `(partition, n_genomes)` where `n_genomes` is
/// the number of genome columns that source contributes. A merged index
/// contributes more than one. The total new columns added to the destination
/// is `sum(n_genomes)`.
///
/// `n_dst_genomes` is the number of genome columns already in the destination
/// matrices (copied from source_0 before this call).
pub fn merge_partition(
&self,
i: usize,
sources: &[(&KmerIndex, usize)],
mode: MergeMode,
n_dst_genomes: usize,
block_bits: u8,
evidence: &IndexMode,
) -> OKIResult<usize> {
let dst_index_dir = self.index_dir(i);
if !dst_index_dir.exists() {
return Ok(0);
}
load_meta(&dst_index_dir)?; // ensure meta.json exists before LayeredMap::open
let dst_map = Arc::new(LayeredMap::<()>::open(&dst_index_dir)?);
let n_dst_layers = dst_map.n_layers();
let n_src_total: usize = sources.iter().map(|(_, n)| *n).sum();
// First merge in presence mode: init presence matrices on existing layers
// (all slots true — every kmer in those layers belongs to genome_0).
if n_dst_genomes == 1 && mode == MergeMode::Presence {
for l in 0..n_dst_layers {
TypedLayer::<()>::init_presence_matrix(
&layer_dir(&dst_index_dir, l),
dst_map.layer(l).n(),
)?;
}
}
// ── Pass 1: pipeline — parallel file read + dst_map filter + graph fill ─
//
// Source : list of unitigs.bin paths (one per source × layer)
// Flat : open file, emit Vec<CanonicalKmer> batches (BeeGFS parallel I/O)
// Transform: filter via dst_map.query() — thread-safe, LayeredMap<()>: Sync
// Sink : push new kmers into GraphDeBruijn (single thread, no locks needed)
// Collect file paths (propagates load_meta errors before the pipeline starts)
let mut unitig_paths: Vec<PathBuf> = Vec::new();
for (src, _) in sources.iter() {
let src_index_dir = src.index_dir(i);
if !src_index_dir.exists() {
continue;
}
let src_meta = load_meta(&src_index_dir)?;
for l in 0..src_meta.n_layers {
let p = layer_dir(&src_index_dir, l).join("unitigs.bin");
if p.exists() {
unitig_paths.push(p);
}
}
}
let n_src_layers = unitig_paths.len();
debug!("partition {i}: de Bruijn graph build start — {n_src_layers} source layer(s)");
const BATCH: usize = 4096;
let n_workers = rayon::current_num_threads().min(16).max(4);
// At most 2 files open simultaneously: keeps n_workers-2 workers free
// for the Transform stage. Each open file monopolises one worker for the
// full duration of its read, so this must stay well below n_workers.
let max_open = 2;
let dst_filter = Arc::clone(&dst_map);
let g = build_graph(
unitig_paths.into_iter(),
move |path: PathBuf, emit: &mut dyn FnMut(Vec<CanonicalKmer>)| -> OKIResult<()> {
let reader = UnitigFileReader::open_sequential(&path)?;
let mut batch: Vec<CanonicalKmer> = Vec::with_capacity(BATCH);
for (kmer, _, _) in reader.iter_indexed_canonical_kmers() {
batch.push(kmer);
if batch.len() == BATCH {
emit(std::mem::replace(&mut batch, Vec::with_capacity(BATCH)));
}
}
if !batch.is_empty() {
emit(batch);
}
Ok(())
},
move |kmer| dst_filter.query(kmer).is_none(),
n_workers,
max_open,
)?;
let any_new = g.len() > 0;
debug!(
"partition {i}: de Bruijn graph done — {} new kmers",
g.len()
);
// Build new layer from de Bruijn graph if there are new kmers.
let new_layer_idx = n_dst_layers;
let new_layer_dir = layer_dir(&dst_index_dir, new_layer_idx);
let n_new = if any_new {
debug!("partition {i}: unitig traversal start — {} nodes", g.len());
let n_nodes = materialize_layer(g, &new_layer_dir, block_bits, evidence)?;
debug!("partition {i}: MPHF build done");
n_nodes
} else {
drop(g);
0
};
let t_open = std::time::Instant::now();
let new_mphf: Option<Arc<MphfOnly>> = if any_new {
Some(Arc::new(MphfOnly::open(&new_layer_dir)?))
} else {
None
};
debug!(
"partition {i}: MPHF open in {:.3}s",
t_open.elapsed().as_secs_f64()
);
// ── Prepare matrix directories for the new layer ──────────────────────
// Absent columns (dst genomes) get an all-zero/false column. Source-genome
// columns are created as mutable builders for pass 2. `new_mb` is kept
// open (not `close`d) until pass 2 has filled every source column, so its
// `n_cols` bookkeeping stays in sync with the columns actually created.
let (new_src_builders, new_mb): (Vec<ColBuilder>, Option<MatrixBuilder>) = if any_new {
let data_dir = match mode {
MergeMode::Presence => new_layer_dir.join("presence"),
MergeMode::Count => new_layer_dir.join("counts"),
};
fs::create_dir_all(&data_dir)?;
let mut mb = MatrixBuilder::new(mode, n_new, &data_dir).map_err(OKIError::Io)?;
for _ in 0..n_dst_genomes {
mb.add_absent_col().map_err(OKIError::Io)?;
}
let cols = (0..n_src_total)
.map(|_| mb.add_col().map_err(OKIError::Io))
.collect::<OKIResult<Vec<_>>>()?;
(cols, Some(mb))
} else {
(vec![], None)
};
let t_builders = std::time::Instant::now();
// Builders for existing layers: n_src_total per layer, resumed from
// each layer's own matrix directory (already holding n_dst_genomes
// columns from a previous merge). Columns land at
// n_dst_genomes .. n_dst_genomes + n_src_total - 1.
let mut exist_mbs: Vec<MatrixBuilder> = Vec::with_capacity(n_dst_layers);
let mut exist_builders: Vec<Vec<ColBuilder>> = Vec::with_capacity(n_dst_layers);
for l in 0..n_dst_layers {
let layer_dir = layer_dir(&dst_index_dir, l);
let data_dir = match mode {
MergeMode::Presence => layer_dir.join("presence"),
MergeMode::Count => layer_dir.join("counts"),
};
let mut mb = MatrixBuilder::resume(mode, &data_dir).map_err(OKIError::Io)?;
let cols = (0..n_src_total)
.map(|_| mb.add_col().map_err(OKIError::Io))
.collect::<OKIResult<Vec<_>>>()?;
exist_mbs.push(mb);
exist_builders.push(cols);
}
debug!(
"partition {i}: builders ready in {:.3}s",
t_builders.elapsed().as_secs_f64()
);
// ── Pass 2: fill builders (pipeline) ─────────────────────────────────
let t_pass2 = std::time::Instant::now();
// Collect source items before the pipeline so load_meta errors propagate
// via ? before any worker thread is spawned.
let mut pass2_items: Vec<(usize, usize, PathBuf)> = Vec::new();
{
let mut col_offset = 0usize;
for (src, src_n) in sources.iter() {
let src_index_dir = src.index_dir(i);
if !src_index_dir.exists() {
col_offset += src_n;
continue;
}
let src_meta = load_meta(&src_index_dir)?;
for l in 0..src_meta.n_layers {
let src_layer_dir = layer_dir(&src_index_dir, l);
if src_layer_dir.join("unitigs.bin").exists() {
pass2_items.push((col_offset, *src_n, src_layer_dir));
}
}
col_offset += src_n;
}
}
enum Pass2Data {
SrcLayer((usize, usize, PathBuf, ThrottleGuard)),
RawBatch((usize, usize, Arc<SrcLayerData>, Vec<CanonicalKmer>)),
WriteBatch(Vec<(Option<usize>, usize, usize, u32)>),
}
let exist_locked: Vec<Vec<Arc<Mutex<ColBuilder>>>> = exist_builders
.into_iter()
.map(|layer| layer.into_iter().map(|b| Arc::new(Mutex::new(b))).collect())
.collect();
let new_locked: Vec<Arc<Mutex<ColBuilder>>> = new_src_builders
.into_iter()
.map(|b| Arc::new(Mutex::new(b)))
.collect();
let exist_sink: Vec<Vec<Arc<Mutex<ColBuilder>>>> = exist_locked
.iter()
.map(|layer| layer.iter().map(Arc::clone).collect())
.collect();
let new_sink: Vec<Arc<Mutex<ColBuilder>>> = new_locked.iter().map(Arc::clone).collect();
let dst_map_t2 = Arc::clone(&dst_map);
let new_mphf_t2 = new_mphf.clone();
let pass2_err: Arc<Mutex<Option<String>>> = Arc::new(Mutex::new(None));
let err_cap2 = Arc::clone(&pass2_err);
let capacity = 2;
let throttled_pass2 = throttle(pass2_items.into_iter(), max_open);
let pipeline2 = Pipeline::new(
make_source!(
Pass2Data,
throttled_pass2.map(|t| {
let (col_offset, src_n, src_layer_dir) = t.item;
(col_offset, src_n, src_layer_dir, t.guard)
}),
SrcLayer
),
vec![
Stage::Flat(Arc::new(
move |data: Pass2Data,
push: &PipelineSender<Result<Pass2Data, PipelineError>>,
delta: &PipelineSender<isize>| {
if let Pass2Data::SrcLayer((col_offset, src_n, src_layer_dir, _guard)) =
data
{
// _guard dropped at end of block, releasing the slot.
let reader = match UnitigFileReader::open_sequential(
&src_layer_dir.join("unitigs.bin"),
) {
Ok(r) => r,
Err(e) => {
*err_cap2.lock().unwrap() = Some(e.to_string());
delta.send(-1).ok();
return;
}
};
let src_data = match SrcLayerData::open(&src_layer_dir, mode) {
Ok(d) => Arc::new(d),
Err(e) => {
*err_cap2.lock().unwrap() = Some(e.to_string());
delta.send(-1).ok();
return;
}
};
const BATCH: usize = 4096;
let mut batch: Vec<CanonicalKmer> = Vec::with_capacity(BATCH);
let mut count: isize = 0;
for (kmer, _, _) in reader.iter_indexed_canonical_kmers() {
batch.push(kmer);
if batch.len() == BATCH {
let b =
std::mem::replace(&mut batch, Vec::with_capacity(BATCH));
push.send(Ok(Pass2Data::RawBatch((
col_offset,
src_n,
Arc::clone(&src_data),
b,
))))
.ok();
count += 1;
}
}
if !batch.is_empty() {
push.send(Ok(Pass2Data::RawBatch((
col_offset, src_n, src_data, batch,
))))
.ok();
count += 1;
}
delta.send(count - 1).ok();
}
},
) as SharedFlatFn<Pass2Data>),
make_transform!(
Pass2Data,
{
move |(col_offset, src_n, src_data, kmers): (
usize,
usize,
Arc<SrcLayerData>,
Vec<CanonicalKmer>,
)|
-> Vec<(Option<usize>, usize, usize, u32)> {
let mut ops: Vec<(Option<usize>, usize, usize, u32)> = Vec::new();
for kmer in kmers {
let values = src_data.lookup(kmer, src_n);
if let Some((dst_layer, hit)) = dst_map_t2.query(kmer) {
for (g, val) in values.into_iter().enumerate() {
ops.push((Some(dst_layer), col_offset + g, hit.slot, val));
}
} else if let Some(ref mphf) = new_mphf_t2 {
let slot = mphf.index(kmer);
for (g, val) in values.into_iter().enumerate() {
ops.push((None, col_offset + g, slot, val));
}
}
}
ops
}
},
RawBatch,
WriteBatch
),
],
make_sink!(
Pass2Data,
{
move |ops: Vec<(Option<usize>, usize, usize, u32)>| {
for (layer_opt, col, slot, val) in ops {
match layer_opt {
Some(l) => exist_sink[l][col].lock().unwrap().set_val(slot, val),
None => new_sink[col].lock().unwrap().set_val(slot, val),
}
}
}
},
WriteBatch
),
);
WorkerPool::new(pipeline2, n_workers, capacity).run();
debug!(
"partition {i}: pass2 pipeline done in {:.3}s",
t_pass2.elapsed().as_secs_f64()
);
if let Some(msg) = Arc::try_unwrap(pass2_err)
.unwrap_or_else(|_| panic!("pass2: pass2_err not uniquely owned"))
.into_inner()
.unwrap_or_else(|e| e.into_inner())
{
return Err(OKIError::InvalidData {
context: "merge pass2",
detail: msg,
});
}
let t_close = std::time::Instant::now();
// ── Close builders and update metadata ────────────────────────────────
for (mb, builders) in exist_mbs.into_iter().zip(exist_locked.into_iter()) {
for b in builders {
Arc::try_unwrap(b)
.unwrap_or_else(|_| panic!("pass2: exist_builder not uniquely owned"))
.into_inner()
.unwrap_or_else(|e| e.into_inner())
.close()?;
}
mb.close().map_err(OKIError::Io)?;
}
for b in new_locked {
Arc::try_unwrap(b)
.unwrap_or_else(|_| panic!("pass2: new_builder not uniquely owned"))
.into_inner()
.unwrap_or_else(|e| e.into_inner())
.close()?;
}
if let Some(mb) = new_mb {
mb.close().map_err(OKIError::Io)?;
let mut part_meta = self.partition_meta(i)?;
part_meta.n_layers = new_layer_idx + 1;
part_meta
.save(&dst_index_dir)?;
}
debug!(
"partition {i}: builders closed in {:.3}s",
t_close.elapsed().as_secs_f64()
);
Ok(n_new)
}
}
@@ -1,94 +0,0 @@
use std::path::Path;
use obicompactvec::{MatrixGroupOps, PersistentBitMatrix, PersistentCompactIntMatrix};
use obikseq::CanonicalKmer;
use obikindex::layer::MphfOnly;
use obikindex::{OKIError, OKIResult};
use super::MergeMode;
// ── SrcLayerData — opened source matrix for pass-2 lookup ─────────────────────
pub enum SrcLayerData {
Presence(MphfOnly, PersistentBitMatrix),
Count(MphfOnly, PersistentCompactIntMatrix),
}
impl SrcLayerData {
pub fn open(layer_dir: &Path, merge_mode: MergeMode) -> OKIResult<Self> {
let counts_dir = layer_dir.join("counts");
match merge_mode {
MergeMode::Presence => {
if counts_dir.exists() && !layer_dir.join("presence").exists() {
let mphf = MphfOnly::open(layer_dir)?;
let mat = PersistentCompactIntMatrix::open(layer_dir).map_err(OKIError::Io)?;
Ok(SrcLayerData::Count(mphf, mat))
} else {
// presence dir exists, or neither exists → Implicit handled by open()
let mphf = MphfOnly::open(layer_dir)?;
let mat = PersistentBitMatrix::open(layer_dir).map_err(OKIError::Io)?;
Ok(SrcLayerData::Presence(mphf, mat))
}
}
MergeMode::Count => {
let mphf = MphfOnly::open(layer_dir)?;
if counts_dir.exists() {
let mat = PersistentCompactIntMatrix::open(layer_dir).map_err(OKIError::Io)?;
Ok(SrcLayerData::Count(mphf, mat))
} else {
// No counts → treat as implicit presence (all 1s)
let mat = PersistentBitMatrix::open(layer_dir).map_err(OKIError::Io)?;
Ok(SrcLayerData::Presence(mphf, mat))
}
}
}
}
/// Return one value per source genome for `kmer`.
/// The caller guarantees `kmer` is in the source MPHF domain.
#[inline]
pub(crate) fn lookup(&self, kmer: CanonicalKmer, n_genomes: usize) -> Vec<u32> {
let mut buf = vec![0u32; n_genomes];
match self {
SrcLayerData::Presence(mphf, mat) => mat.fill_row(mphf.index(kmer), &mut buf),
SrcLayerData::Count(mphf, mat) => mat.fill_row(mphf.index(kmer), &mut buf),
}
buf
}
pub fn n_slots(&self) -> usize {
match self {
SrcLayerData::Presence(_, mat) => mat.n(),
SrcLayerData::Count(_, mat) => mat.n(),
}
}
/// MPHF lookup: returns the slot index for `kmer` (kmer must be in the domain).
#[inline]
pub fn slot(&self, kmer: CanonicalKmer) -> usize {
match self {
SrcLayerData::Presence(mphf, _) => mphf.index(kmer),
SrcLayerData::Count(mphf, _) => mphf.index(kmer),
}
}
/// Row lookup by slot index, bypassing the MPHF.
#[inline]
pub fn fill_row_by_slot(&self, slot: usize, n_genomes: usize) -> Vec<u32> {
let mut buf = vec![0u32; n_genomes];
match self {
SrcLayerData::Presence(_, mat) => mat.fill_row(slot, &mut buf),
SrcLayerData::Count(_, mat) => mat.fill_row(slot, &mut buf),
}
buf
}
/// Call `f` with a reference to the underlying matrix as `&dyn MatrixGroupOps`.
pub fn with_matrix<R>(&self, f: impl FnOnce(&dyn MatrixGroupOps) -> R) -> R {
match self {
SrcLayerData::Presence(_, mat) => f(mat),
SrcLayerData::Count(_, mat) => f(mat),
}
}
}
@@ -0,0 +1,207 @@
//! The full, unsampled entropy scan behind `--shannon` and
//! `--shannon-summary`: every non-monomorphic minorant of every cached layer,
//! resolved once, with both entropies and the genome counts they were computed
//! over. [`scan_family_entropies`] is the one traversal; the two outputs differ
//! only in what they do with each [`FamilyEntropyRow`] — one row per family
//! (huge on a large index), or counts per bin ([`EntropySummary`], a few
//! hundred thousand rows at most whatever the index size).
use std::collections::{BTreeMap, HashSet};
use std::io::{self, Write};
use obikidxcache::index_cache::IndexCache;
use obikindex::{OKIError, OKIResult};
use obisys::progress_bar;
use super::entropy::{family_entropy, family_entropy_4};
use super::family_scan::{Selection, scan_layer_families};
use super::is_fast_mode;
use super::minorant_selection::non_monomorphic_reindex_layer;
use crate::siblings::extensions::SiblingBuilder;
/// One variable family, as `--shannon` reports it.
pub(crate) struct FamilyEntropyRow {
pub partition: usize,
pub layer: usize,
pub family_idx: usize,
pub family_size: usize,
/// Entropy over the 15 non-empty states, every genome carrying the family.
pub entropy15: f64,
/// Entropy over the 4 bases, genomes carrying exactly one member only;
/// `0.0` (and `n_unambiguous == 0`) when there is none.
pub entropy4: f64,
pub n_present: usize,
pub n_unambiguous: usize,
}
/// Calls `visit` once per non-monomorphic family of the index, layer by layer.
/// Needs only the sibling annex (`non_monomorphic_reindex_layer`): it
/// recomputes every value, so it neither reads nor requires the entropy,
/// quartet, overlap or pair-weight annexes. Bounded to non-monomorphic
/// minorants — monomorphism is already known from the annex bits alone (~98%
/// of a layer's minorants, measured on real data), so there is no reason to
/// pay `scan_layer_families`'s expensive cross-partition resolution for
/// families that would be skipped right away. A failing `visit` stops the scan.
pub(crate) fn scan_family_entropies(
cache: &IndexCache,
mut visit: impl FnMut(&FamilyEntropyRow) -> io::Result<()>,
) -> OKIResult<()> {
cache.ensure_sibling_annex()?;
let n_genomes = cache.meta().genomes().len();
let fast_mode = is_fast_mode(cache);
let pb = progress_bar("shannon_entropy", cache.n_partition() as u64, "partitions");
for partition in cache.partitions() {
let n_layer = cache.n_layer(partition).unwrap_or(0);
for layer_idx in 0..n_layer {
let layer = cache
.get_layer(partition, layer_idx)
.expect("cache-consistent: n_layer(partition) already bounds layer_idx");
let reindex = non_monomorphic_reindex_layer(layer.dir())?;
let eligible: HashSet<usize> = reindex.keys().copied().collect();
let mut failure = None;
scan_layer_families(
cache,
partition,
layer_idx,
n_genomes,
fast_mode,
&Selection::Some(&eligible),
|family_idx, mask, genome_mask| {
if failure.is_some() {
return;
}
debug_assert!(
mask.family_size() >= 2,
"Selection::Some(eligible) must only visit non-monomorphic minorants"
);
let Some((entropy15, n_present)) = family_entropy(genome_mask) else {
return;
};
let (entropy4, n_unambiguous) =
family_entropy_4(genome_mask).unwrap_or((0.0, 0));
let row = FamilyEntropyRow {
partition,
layer: layer_idx,
family_idx,
family_size: mask.family_size() as usize,
entropy15,
entropy4,
n_present,
n_unambiguous,
};
if let Err(e) = visit(&row) {
failure = Some(e);
}
},
)?;
if let Some(e) = failure {
return Err(OKIError::Io(e));
}
}
pb.inc(1);
}
pb.finish_and_clear();
Ok(())
}
/// Width, in bits, of an entropy bin of [`EntropySummary`].
pub(crate) const ENTROPY_BIN_WIDTH: f64 = 0.05;
/// Which entropy a summary line is about.
#[derive(Clone, Copy, Debug, PartialEq, Eq, PartialOrd, Ord)]
enum Which {
Entropy15,
Entropy4,
}
/// Number of families per (entropy, genomes present, genomes without
/// ambiguity, entropy bin) — what `--shannon` rows reduce to once the
/// question is a distribution (entropy against the number of carriers, how
/// many carriers `--no-ambiguity` loses) and not a particular family.
#[derive(Default)]
pub(crate) struct EntropySummary {
counts: BTreeMap<(Which, usize, usize, usize), u64>,
}
impl EntropySummary {
fn bin(entropy: f64) -> usize {
(entropy / ENTROPY_BIN_WIDTH).floor() as usize
}
pub(crate) fn add(&mut self, row: &FamilyEntropyRow) {
*self
.counts
.entry((Which::Entropy15, row.n_present, row.n_unambiguous, Self::bin(row.entropy15)))
.or_default() += 1;
// `entropy4` is not defined without a genome carrying exactly one member
if row.n_unambiguous > 0 {
*self
.counts
.entry((Which::Entropy4, row.n_present, row.n_unambiguous, Self::bin(row.entropy4)))
.or_default() += 1;
}
}
pub(crate) fn write(&self, w: &mut impl Write) -> io::Result<()> {
writeln!(w, "entropy,n_genomes_present,n_genomes_unambiguous,bin_start,n_families")?;
for (&(which, present, unambiguous, bin), &count) in &self.counts {
let name = match which {
Which::Entropy15 => "entropy15",
Which::Entropy4 => "entropy4",
};
writeln!(
w,
"{name},{present},{unambiguous},{:.2},{count}",
bin as f64 * ENTROPY_BIN_WIDTH
)?;
}
Ok(())
}
}
#[cfg(test)]
mod tests {
use super::*;
fn row(entropy15: f64, entropy4: f64, n_present: usize, n_unambiguous: usize) -> FamilyEntropyRow {
FamilyEntropyRow {
partition: 0,
layer: 0,
family_idx: 0,
family_size: 2,
entropy15,
entropy4,
n_present,
n_unambiguous,
}
}
fn written(summary: &EntropySummary) -> Vec<String> {
let mut buf = Vec::new();
summary.write(&mut buf).unwrap();
String::from_utf8(buf).unwrap().lines().map(str::to_owned).collect()
}
#[test]
fn families_are_counted_per_carrier_numbers_and_entropy_bin() {
let mut s = EntropySummary::default();
s.add(&row(1.0, 1.0, 4, 4));
s.add(&row(1.02, 1.02, 4, 4)); // same 0.05 bin as 1.0
s.add(&row(0.2, 0.2, 4, 4));
let lines = written(&s);
assert_eq!(lines[0], "entropy,n_genomes_present,n_genomes_unambiguous,bin_start,n_families");
assert!(lines.contains(&"entropy15,4,4,1.00,2".to_string()));
assert!(lines.contains(&"entropy15,4,4,0.20,1".to_string()));
assert!(lines.contains(&"entropy4,4,4,1.00,2".to_string()));
}
#[test]
fn entropy4_is_left_out_when_no_genome_has_a_single_member() {
let mut s = EntropySummary::default();
s.add(&row(0.0, 0.0, 1, 0));
let lines = written(&s);
assert!(lines.contains(&"entropy15,1,0,0.00,1".to_string()));
assert!(!lines.iter().any(|l| l.starts_with("entropy4")));
}
}
@@ -22,7 +22,9 @@ use super::masking::is_quartet_informative;
use super::minorant_selection::non_monomorphic_reindex_layer;
use super::pair_weight::pair_balance_weight;
use crate::siblings::quartet_annex::quartet_annex_file_name;
use crate::siblings::{FamilyOverlapAccumulator, FamilyScalarAnnexBuilder, ScalarKind};
use crate::siblings::{
FamilyOverlapAccumulator, FamilyScalarAnnexBuilder, ScalarKind, family_overlap_file_name,
};
/// Which annexes a caller needs, each listed once per `--no-ambiguity`
/// counting convention (`false` = genomes carrying several members count as a
@@ -50,6 +52,24 @@ impl AnnexRequest {
dedup(self.overlap.iter().chain(&self.pair_weight).copied())
}
/// Whether every annex this request asks for already exists — the same
/// per-file existence test the builders apply, so `true` means they would
/// find nothing to build and no lock is needed. Cheap (no file is opened).
pub(crate) fn is_satisfied(&self, cache: &IndexCache) -> bool {
let overlap = self
.overlap_conventions()
.into_iter()
.all(|nb| cache.index().dir().join(family_overlap_file_name(nb)).exists());
let pair_weight = self.pair_weight_conventions();
overlap
&& cache.iter_indexed().all(|(_, _, layer)| {
let has = |name: &str| layer.dir().join(name).exists();
self.entropy.iter().all(|&nb| has(ScalarKind::Entropy.file_name(nb)))
&& self.quartet.iter().all(|&nb| has(quartet_annex_file_name(nb)))
&& pair_weight.iter().all(|&nb| has(ScalarKind::PairWeight.file_name(nb)))
})
}
/// The pair-weight conventions requested, each once.
pub(crate) fn pair_weight_conventions(&self) -> Vec<bool> {
dedup(self.pair_weight.iter().copied())
@@ -54,7 +54,7 @@ use obipipeline::ThrottleGuard;
use crate::siblings::helpers::central_base;
use crate::siblings::iter::{SiblingEntry, SiblingLayerExt};
use crate::siblings::{ANNEX_FILE_NAME, FamilyMask, SiblingAnnex};
use crate::siblings::{FamilyMask, SiblingAnnex};
/// Target memory budget for one batch's `genome_mask` allocation
/// (`family_batch_size(n_genomes) * n_genomes` bytes, see below) — bounding
@@ -180,7 +180,7 @@ pub(crate) fn scan_layer_families(
})?;
let n_cols = mat.n_cols().min(n_genomes);
let layer_dir: &Path = mat.dir();
let annex = Arc::new(SiblingAnnex::open(&layer_dir.join(ANNEX_FILE_NAME)).map_err(OKIError::Io)?);
let annex = Arc::new(SiblingAnnex::open_in(layer_dir).map_err(OKIError::Io)?);
let ctx = Arc::new(LayerCtx { mat, n_parts, n_genomes, n_cols, k });
@@ -229,6 +229,9 @@ pub(crate) fn scan_layer_families(
let mut genome_mask = vec![0u8; n * ctx.n_genomes];
let mut outgoing: Vec<Vec<(CanonicalKmer, usize, u8, u8)>> =
(0..ctx.n_parts).map(|_| Vec::new()).collect();
// Batch positions of the families `selected` keeps — the only
// ones pass 2 reads the matrix for.
let mut picked: Vec<usize> = Vec::new();
// Pass 1: cheap, no matrix access — own base and this
// batch's cross-partition queries. Kmer and mask already in
@@ -239,8 +242,9 @@ pub(crate) fn scan_layer_families(
// hands back each one's annex-recorded layer alongside.
// Families outside `selected` (sampling only) still get a
// mask/base entry — pass 2 indexes uniformly by `i` — but
// never an outgoing query: that's the expensive part a
// sampled run exists to avoid paying for every family.
// are neither queried across partitions nor read from the
// matrix in pass 2: that's the expensive part a sampled run
// exists to avoid paying for every family.
for (i, entry) in batch.entries.iter().enumerate() {
let (kmer, mask) = (entry.kmer, entry.mask);
masks.push(mask);
@@ -250,6 +254,7 @@ pub(crate) fn scan_layer_families(
if !is_selected(&worker_selected, family_idx) {
continue;
}
picked.push(i);
for (member, layer) in mask.family_members(kmer, ctx.k) {
if member == kmer {
continue; // local — resolved below straight from `mat`, no lookup
@@ -269,14 +274,22 @@ pub(crate) fn scan_layer_families(
// each entry's kmer is mapped to its slot via `hash_batch`
// — a pure MPHF lookup, no evidence check, since these are
// this layer's own kmers, known members by construction.
let kmers: Vec<CanonicalKmer> = batch.entries.iter().map(|e| e.kmer).collect();
let slots = ctx.mat.hash_batch(&kmers);
let mut carries: Vec<Vec<bool>> = (0..ctx.n_cols).map(|_| Vec::new()).collect();
ctx.mat.fill_sub_matrix_carries(&slots, &mut carries);
for (g, col) in carries.iter().enumerate() {
for (i, &carries_it) in col.iter().enumerate() {
if carries_it {
genome_mask[i * ctx.n_genomes + g] |= 1 << bases[i];
// Only the `picked` families are read: a sampled or
// annex-restricted scan visits a handful of a layer's
// minorants, and a full-batch matrix pass would cost as much
// as `Selection::All`. `carries[g][j]` belongs to batch
// position `picked[j]`.
if !picked.is_empty() {
let kmers: Vec<CanonicalKmer> =
picked.iter().map(|&i| batch.entries[i].kmer).collect();
let slots = ctx.mat.hash_batch(&kmers);
let mut carries: Vec<Vec<bool>> = (0..ctx.n_cols).map(|_| Vec::new()).collect();
ctx.mat.fill_sub_matrix_carries(&slots, &mut carries);
for (g, col) in carries.iter().enumerate() {
for (&i, &carries_it) in picked.iter().zip(col) {
if carries_it {
genome_mask[i * ctx.n_genomes + g] |= 1 << bases[i];
}
}
}
}
@@ -72,7 +72,7 @@ pub(crate) fn survives_masking(masked: &[Option<u8>], excluded: &[bool]) -> bool
/// genomes. That is the pattern that can separate two pairs of genomes from
/// each other (`AB|CD`); a singleton variant, or a family carried by fewer
/// than four genomes, fits every topology and informs none.
fn has_bipartition(counts: &[u32; 16]) -> bool {
pub(crate) fn has_bipartition(counts: &[u32; 16]) -> bool {
counts[1..].iter().filter(|&&c| c >= 2).count() >= 2
}
@@ -11,7 +11,7 @@ use std::path::Path;
use obicompactvec::TempBitVecBuilder;
use obikindex::{OKIError, OKIResult};
use crate::siblings::{ANNEX_FILE_NAME, FamilyMask, FamilyScalarAnnex, ScalarKind, SiblingAnnex};
use crate::siblings::{FamilyMask, FamilyScalarAnnex, ScalarKind, SiblingAnnex};
pub(crate) fn is_non_monomorphic_minorant(mask: FamilyMask) -> bool {
mask.is_minorant() && mask.siblings() >= 1
@@ -31,7 +31,7 @@ pub(crate) fn is_non_monomorphic_minorant(mask: FamilyMask) -> bool {
/// the `Selection` a caller needs to bound `scan_layer_families` to just
/// these families.
pub(crate) fn non_monomorphic_reindex_layer(layer_dir: &Path) -> OKIResult<HashMap<usize, usize>> {
let annex = SiblingAnnex::open(&layer_dir.join(ANNEX_FILE_NAME)).map_err(OKIError::Io)?;
let annex = SiblingAnnex::open_in(layer_dir).map_err(OKIError::Io)?;
let mut reindex = HashMap::new();
let mut family_idx = 0usize;
for slot in 0..annex.len() {
@@ -62,7 +62,7 @@ pub(crate) fn non_monomorphic_reindex_layer(layer_dir: &Path) -> OKIResult<HashM
/// first counts this layer's total minorants (needed up front to size the
/// bitset), the second sets the eligible bits.
pub(crate) fn eligible_bitset_layer(layer_dir: &Path) -> OKIResult<TempBitVecBuilder> {
let annex = SiblingAnnex::open(&layer_dir.join(ANNEX_FILE_NAME)).map_err(OKIError::Io)?;
let annex = SiblingAnnex::open_in(layer_dir).map_err(OKIError::Io)?;
let n_minorants = (0..annex.len())
.filter(|&slot| annex.get(slot).is_some_and(|m| m.is_minorant()))
@@ -103,7 +103,7 @@ pub(crate) fn iter_full_scalar(
kind: ScalarKind,
no_ambiguity: bool,
) -> OKIResult<impl Iterator<Item = f32> + use<>> {
let annex = SiblingAnnex::open(&layer_dir.join(ANNEX_FILE_NAME)).map_err(OKIError::Io)?;
let annex = SiblingAnnex::open_in(layer_dir).map_err(OKIError::Io)?;
let scalars =
FamilyScalarAnnex::open(&layer_dir.join(kind.file_name(no_ambiguity)), kind).map_err(OKIError::Io)?;
if scalars.no_ambiguity() != no_ambiguity {
+5 -4
View File
@@ -17,6 +17,7 @@ mod alignment;
mod annex;
mod cardcomp;
mod entropy;
mod entropy_scan;
mod family_annexes;
mod family_scan;
mod masking;
@@ -24,6 +25,7 @@ mod minorant_selection;
mod pair_weight;
mod pairwise;
mod quartet;
mod sample_stats;
mod sankoff;
mod scan;
mod snp_distance;
@@ -41,17 +43,16 @@ pub use sankoff::SankoffBundle;
pub use scan::{LayerReducer, sampled_scan};
pub use snp_distance::{GammaShape, SnpDistanceKind, distance_matrix};
pub use stats::SiblingAnnexStats;
pub use subsample::{EntropyBias, SamplingSpec, SurvivingFamily};
pub use sample_stats::{SampleStats, SampleStatsReducer};
pub use subsample::{EntropyBias, LayerReport, SamplePlan, SamplingSpec, SurvivingFamily};
pub(crate) use annex::build_layer_sibling_annex;
pub(crate) use entropy::{family_entropy, family_entropy_4};
pub(crate) use entropy_scan::{EntropySummary, scan_family_entropies};
pub(crate) use family_annexes::{
AnnexRequest, ensure_layer_family_annexes, ensure_layer_pair_weight_annexes,
};
pub(crate) use pair_weight::inverse_overlap_matrix;
pub(crate) use quartet::restrict_to_quartet;
pub(crate) use family_scan::{Selection, scan_layer_families};
pub(crate) use minorant_selection::non_monomorphic_reindex_layer;
pub(crate) use stats::{sibling_annex_stats, sibling_family_size_histogram};
/// Whether every layer number in `cache` fits in a `FamilyMask` field
@@ -8,7 +8,7 @@ use obikindex::{OKIError, OKIResult};
use super::minorant_selection::is_non_monomorphic_minorant;
use crate::siblings::quartet_annex::quartet_annex_file_name;
use crate::siblings::{ANNEX_FILE_NAME, SiblingAnnex};
use crate::siblings::SiblingAnnex;
/// Clears, in `eligible` (indexed by `family_idx`, monomorphic minorants
/// included), every family the quartet annex flags as not quartet-informative
@@ -22,7 +22,7 @@ pub(crate) fn restrict_to_quartet(
layer_dir: &Path,
no_ambiguity: bool,
) -> OKIResult<()> {
let annex = SiblingAnnex::open(&layer_dir.join(ANNEX_FILE_NAME)).map_err(OKIError::Io)?;
let annex = SiblingAnnex::open_in(layer_dir).map_err(OKIError::Io)?;
let flags = PersistentBitVec::open(&layer_dir.join(quartet_annex_file_name(no_ambiguity)))
.map_err(OKIError::Io)?;
@@ -0,0 +1,164 @@
//! `--sample-stats`: what a draw actually contains, accumulated as a
//! [`LayerReducer`] over [`super::scan::sampled_scan`]'s batches, side by side
//! with the alignment, from the very same families — per genome (how many
//! sampled sites it has a call in, how many of them can resolve a
//! bipartition), per number of carriers (how many sampled sites, how many
//! informative), and per pair of genomes (how many sampled sites both have a
//! call in: the sample's own family-overlap matrix, to set against the index's).
//!
//! A *call* is a masked state other than the empty one (`0`): absence is never
//! a call, with or without `--free-loss`. A site is *informative* when its
//! carriers show a bipartition (`masking::has_bipartition`).
use super::masking::has_bipartition;
use super::scan::LayerReducer;
use super::subsample::SurvivingFamily;
use crate::siblings::helpers::triangle_index;
/// The statistics of one draw. `pub`: part of the public signature of the
/// `--sample-stats` output.
#[derive(Clone, Debug)]
pub struct SampleStats {
pub n_genomes: usize,
/// Number of sampled sites.
pub n_sites: u64,
/// Per genome: sampled sites it has a call in.
pub calls: Vec<u64>,
/// Per genome: informative sampled sites in which its state is one of the
/// two or more carried by at least two genomes — those where it takes
/// part in the bipartition.
pub informative_calls: Vec<u64>,
/// Index = number of carriers (`0..=n_genomes`): sampled sites.
pub sites_by_carriers: Vec<u64>,
/// Same, informative sites only.
pub informative_by_carriers: Vec<u64>,
/// Sampled sites both genomes of a pair have a call in, upper triangle in
/// the family-overlap annex's own order — read it with
/// [`overlap`](Self::overlap).
overlap: Vec<u64>,
}
impl SampleStats {
/// Sampled sites genomes `i` and `j` both have a call in (`0` for `i == j`,
/// as in `FamilyOverlap::get`).
pub fn overlap(&self, i: usize, j: usize) -> u64 {
if i == j {
0
} else {
self.overlap[triangle_index(self.n_genomes, i, j)]
}
}
}
/// Accumulates a [`SampleStats`] over a [`super::scan::sampled_scan`] pass.
/// `excluded[g]` genomes are not counted anywhere, as in the alignment.
pub struct SampleStatsReducer {
excluded: Vec<bool>,
stats: SampleStats,
}
impl SampleStatsReducer {
pub fn new(n_genomes: usize, excluded: &[bool]) -> Self {
let n_pairs = n_genomes.saturating_sub(1) * n_genomes / 2;
Self {
excluded: (0..n_genomes).map(|g| excluded.get(g).copied().unwrap_or(false)).collect(),
stats: SampleStats {
n_genomes,
n_sites: 0,
calls: vec![0; n_genomes],
informative_calls: vec![0; n_genomes],
sites_by_carriers: vec![0; n_genomes + 1],
informative_by_carriers: vec![0; n_genomes + 1],
overlap: vec![0; n_pairs],
},
}
}
pub fn finish(self) -> SampleStats {
self.stats
}
/// One sampled site, from its per-genome masked states.
fn add_site(&mut self, masked: &[Option<u8>]) {
let n = self.stats.n_genomes;
let mut counts = [0u32; 16];
let mut carriers: Vec<usize> = Vec::new();
for (g, &state) in masked.iter().enumerate() {
if let Some(state) = state.filter(|&s| s != 0 && !self.excluded[g]) {
counts[(state & 0b1111) as usize] += 1;
carriers.push(g);
}
}
let informative = has_bipartition(&counts);
self.stats.n_sites += 1;
self.stats.sites_by_carriers[carriers.len()] += 1;
if informative {
self.stats.informative_by_carriers[carriers.len()] += 1;
}
for (a, &i) in carriers.iter().enumerate() {
self.stats.calls[i] += 1;
if informative && counts[(masked[i].unwrap() & 0b1111) as usize] >= 2 {
self.stats.informative_calls[i] += 1;
}
for &j in &carriers[a + 1..] {
self.stats.overlap[triangle_index(n, i, j)] += 1;
}
}
}
}
impl LayerReducer for SampleStatsReducer {
fn reduce(&mut self, _partition: usize, _layer: usize, survivors: &[SurvivingFamily]) {
for family in survivors {
self.add_site(&family.masked);
}
}
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn counts_calls_carriers_and_pairs() {
let mut r = SampleStatsReducer::new(5, &[]);
// informative: A A G G, genome 4 absent (state 0)
r.add_site(&[Some(1), Some(1), Some(4), Some(4), Some(0)]);
// not informative: three A and one lone G, genome 4 missing
r.add_site(&[Some(1), Some(1), Some(1), Some(4), None]);
let s = r.finish();
assert_eq!(s.n_sites, 2);
assert_eq!(s.calls, vec![2, 2, 2, 2, 0]);
assert_eq!(s.sites_by_carriers[4], 2);
assert_eq!(s.informative_by_carriers[4], 1);
// genome 3 is in the G pair of the informative site only
assert_eq!(s.informative_calls, vec![1, 1, 1, 1, 0]);
assert_eq!(s.overlap(0, 1), 2);
assert_eq!(s.overlap(2, 3), 2);
assert_eq!(s.overlap(0, 4), 0);
assert_eq!(s.overlap(1, 1), 0);
assert_eq!(s.overlap(3, 0), s.overlap(0, 3));
}
#[test]
fn excluded_genomes_count_nowhere() {
let mut r = SampleStatsReducer::new(4, &[false, false, false, true]);
r.add_site(&[Some(1), Some(1), Some(4), Some(4)]);
let s = r.finish();
// without genome 3 the G state has a single carrier: not informative
assert_eq!(s.calls, vec![1, 1, 1, 0]);
assert_eq!(s.sites_by_carriers[3], 1);
assert_eq!(s.informative_by_carriers[3], 0);
assert_eq!(s.overlap(2, 3), 0);
}
#[test]
fn genome_with_a_lone_state_takes_no_part_in_the_bipartition() {
let mut r = SampleStatsReducer::new(5, &[]);
// A A C C G: genome 4 is the lone G
r.add_site(&[Some(1), Some(1), Some(2), Some(2), Some(4)]);
let s = r.finish();
assert_eq!(s.informative_calls, vec![1, 1, 1, 1, 0]);
assert_eq!(s.calls[4], 1);
}
}
@@ -14,7 +14,7 @@ use obikidxcache::index_cache::IndexCache;
use obikindex::OKIResult;
use super::stats::sibling_family_size_histogram;
use super::subsample::{SamplingSpec, SurvivingFamily, sample_index};
use super::subsample::{SamplePlan, SamplingSpec, SurvivingFamily, sample_index};
/// One accumulator fed every layer's [`SurvivingFamily`] batch during a
/// [`sampled_scan`] pass — self-contained (no sampling/masking logic of
@@ -47,22 +47,22 @@ fn resolve_target(cache: &IndexCache, n: Option<usize>) -> OKIResult<usize> {
/// own docs and [`SamplingSpec`] for what each controls), feeding every
/// surviving family of every visited layer to every reducer in
/// `reducers`, in the same order, before moving to the next layer. A
/// no-op (not an error) if the resolved target is `0`.
/// no-op (not an error) if the resolved target is `0`. Returns how the draw
/// was spread over the layers ([`SamplePlan`]).
pub fn sampled_scan(
cache: &IndexCache,
n: Option<usize>,
spec: SamplingSpec,
excluded: &[bool],
reducers: &mut [&mut dyn LayerReducer],
) -> OKIResult<()> {
) -> OKIResult<SamplePlan> {
let target = resolve_target(cache, n)?;
if target == 0 {
return Ok(());
return Ok(SamplePlan::default());
}
sample_index(cache, target, spec, excluded, 0, |partition, layer, _raw_index, survivors| {
for reducer in reducers.iter_mut() {
reducer.reduce(partition, layer, &survivors);
}
})?;
Ok(())
})
}
+9 -16
View File
@@ -9,11 +9,12 @@
//! already stored — no re-derivation, no lookup).
use obikidxcache::index_cache::IndexCache;
use obikindex::{OKIError, OKIResult};
use obikindex::OKIResult;
use super::family_scan::{Selection, scan_layer_families};
use super::is_fast_mode;
use crate::siblings::{ANNEX_FILE_NAME, SiblingAnnex};
use crate::siblings::SiblingAnnex;
use crate::siblings::extensions::SiblingBuilder;
/// Layers scanned at once per [`sibling_family_size_histogram`]'s
/// `par_map_reduce` call — same bounding reasoning as
@@ -45,26 +46,17 @@ pub struct SiblingAnnexStats {
/// on top of that would fight it for cache/mmap locality) — this function
/// has no such inner parallelism to protect.
///
/// Existence of every cached layer's annex file is checked up front, before
/// the parallel map runs, so a missing one is reported as an error rather
/// than silently read back as zero.
/// A missing annex is built up front (`ensure_sibling_annex`), before the
/// parallel map runs, so none is silently read back as zero.
pub(crate) fn sibling_family_size_histogram(cache: &IndexCache) -> OKIResult<[u64; 4]> {
for (part, l, layer) in cache.iter_indexed() {
let path = layer.dir().join(ANNEX_FILE_NAME);
if !path.exists() {
return Err(OKIError::InvalidInput(format!(
"no sibling annex at {} (partition {part}, layer {l}) — run --sibling-annex first",
path.display()
)));
}
}
cache.ensure_sibling_annex()?;
Ok(cache.par_map_reduce(
MAX_CONCURRENT_LAYERS,
|| [0u64; 4],
|layer| {
let annex = SiblingAnnex::open(&layer.dir().join(ANNEX_FILE_NAME))
.expect("existence already checked before this parallel map ran");
let annex = SiblingAnnex::open_in(layer.dir())
.expect("ensure_sibling_annex already made every layer's annex exist");
let mut counts = [0u64; 4];
for slot in 0..annex.len() {
let Some(mask) = annex.get(slot) else { continue };
@@ -95,6 +87,7 @@ pub(crate) fn sibling_family_size_histogram(cache: &IndexCache) -> OKIResult<[u6
/// same reasoning `SiblingExt::build_sibling_annex`/`shannon_entropy_csv`/
/// `algorithms::scan::sampled_scan` already follow).
pub(crate) fn sibling_annex_stats(cache: &IndexCache) -> OKIResult<SiblingAnnexStats> {
cache.ensure_sibling_annex()?;
let n_genomes = cache.meta().genomes().len();
let fast_mode = is_fast_mode(cache);
@@ -49,6 +49,7 @@ use std::sync::Arc;
use obicompactvec::TempBitVecBuilder;
use obikidxcache::index_cache::IndexCache;
use obikindex::OKIResult;
use obisys::progress_bar;
use rand::RngExt;
use super::family_scan::{Selection, scan_layer_families};
@@ -115,6 +116,41 @@ pub struct SamplingSpec {
pub entropy_bias: Option<EntropyBias>,
}
/// What one layer contributed to a draw — one row of `--sample-plan`.
#[derive(Clone, Debug)]
pub struct LayerReport {
pub partition: usize,
pub layer: usize,
/// Families eligible when the draw starts (non-monomorphic, and
/// quartet-informative with `--quartet`).
pub eligible: u64,
/// Their total sampling weight: their number, or with `--pair-balance` the
/// sum of their pair weights.
pub mass: f64,
/// The layer's share of the target.
pub quota: usize,
/// Families drawn and resolved.
pub resolved: usize,
/// Of those, the ones that survived masking (and `--quartet`'s exact check).
pub kept: usize,
}
/// How a draw was spread over the layers of the index.
#[derive(Clone, Debug, Default)]
pub struct SamplePlan {
/// The number of sites asked for.
pub target: usize,
/// One report per layer, in the order the draw visited them.
pub layers: Vec<LayerReport>,
}
impl SamplePlan {
/// Sites actually kept, all layers together.
pub fn kept(&self) -> usize {
self.layers.iter().map(|layer| layer.kept).sum()
}
}
/// [`iter_full_scalar`] of `kind` (entropy, pair weight), rotated to start at
/// `start` and wrap once back to it — the same `family_idx` order
/// [`sample_layer`]'s own circular index sequence walks, so the streams can be
@@ -143,6 +179,8 @@ struct LayerPlan {
/// One bit per minorant (`family_idx` order): still eligible to be drawn.
eligible: TempBitVecBuilder,
n_minorants: usize,
/// How many of those bits are set when the draw starts.
eligible_count: u64,
/// Total sampling weight of the eligible families: their number, or, with
/// `--pair-balance`, the sum of their pair weights. A layer's quota is its
/// share of this mass over every layer, so a layer rich in families that
@@ -191,10 +229,10 @@ struct LayerPlan {
/// the seam a future resume mechanism would hook into, rather than being
/// ripped out and re-added later.
///
/// Returns the actual number of *newly processed* sites kept (not
/// counting whatever `skip_layers` layers already contributed on an
/// earlier, interrupted call) — may be less than `n` (see the module
/// docs).
/// Returns how the draw was spread over the layers ([`SamplePlan`]): one
/// report per layer, with its quota and what it actually kept (nothing for the
/// `skip_layers` layers an earlier, interrupted call already covered). The
/// sum of what was kept may be less than `n` (see the module docs).
pub(crate) fn sample_index(
cache: &IndexCache,
n: usize,
@@ -202,7 +240,10 @@ pub(crate) fn sample_index(
excluded: &[bool],
skip_layers: usize,
mut on_layer: impl FnMut(usize, usize, usize, Arc<Vec<SurvivingFamily>>),
) -> OKIResult<usize> {
) -> OKIResult<SamplePlan> {
// Every selection below reads the sibling annex (`eligible_bitset_layer`),
// whichever derived annexes the options ask for.
cache.ensure_sibling_annex()?;
let n_genomes = cache.meta().genomes().len();
let fast_mode = is_fast_mode(cache);
@@ -226,6 +267,7 @@ pub(crate) fn sample_index(
// — needed before any layer's quota can be computed (each layer's share
// depends on every other layer's own mass).
let mut plans: Vec<LayerPlan> = Vec::new();
let pb = progress_bar("sampling_plan", cache.n_partition() as u64, "partitions");
for partition in cache.partitions() {
let n_layer = cache.n_layer(partition).unwrap_or(0);
for layer_idx in 0..n_layer {
@@ -237,6 +279,7 @@ pub(crate) fn sample_index(
restrict_to_quartet(&mut eligible, layer.dir(), spec.no_ambiguity)?;
}
let n_minorants = eligible.view().len();
let eligible_count = eligible.view().count_ones();
let mass = if spec.pair_balance {
iter_full_scalar(layer.dir(), ScalarKind::PairWeight, spec.no_ambiguity)?
.enumerate()
@@ -244,15 +287,25 @@ pub(crate) fn sample_index(
.map(|(_, weight)| f64::from(weight))
.sum()
} else {
eligible.view().count_ones() as f64
eligible_count as f64
};
plans.push(LayerPlan { partition, layer: layer_idx, eligible, n_minorants, mass });
plans.push(LayerPlan {
partition,
layer: layer_idx,
eligible,
n_minorants,
eligible_count,
mass,
});
}
pb.inc(1);
}
pb.finish_and_clear();
let total_mass: f64 = plans.iter().map(|plan| plan.mass).sum();
let mut report = SamplePlan { target: n, layers: Vec::with_capacity(plans.len()) };
if total_mass <= 0.0 {
return Ok(0);
return Ok(report);
}
// Quotas by cumulative rounding: layer `i` is allotted
@@ -260,34 +313,47 @@ pub(crate) fn sample_index(
// were allotted`, so the quotas sum to exactly `n`. Rounding each layer's
// share down on its own would lose up to one site per layer — all of
// them, when `n` is smaller than the number of layers.
let mut total_kept = 0usize;
let mut cumulative_mass = 0.0f64;
let mut allotted = 0usize;
let pb = progress_bar("sampling", plans.len() as u64, "layers");
for (raw_index, mut plan) in plans.into_iter().enumerate() {
cumulative_mass += plan.mass;
let cumulative_quota = (n as f64 * cumulative_mass / total_mass) as usize;
let quota = cumulative_quota.saturating_sub(allotted);
allotted = allotted.max(cumulative_quota);
if raw_index < skip_layers || plan.mass <= 0.0 || quota == 0 {
continue;
}
total_kept += sample_layer(
cache,
raw_index,
&mut plan,
let (resolved, kept) = if raw_index < skip_layers || plan.mass <= 0.0 || quota == 0 {
(0, 0)
} else {
sample_layer(
cache,
raw_index,
&mut plan,
quota,
n_genomes,
fast_mode,
spec,
excluded,
&mut on_layer,
)?
};
report.layers.push(LayerReport {
partition: plan.partition,
layer: plan.layer,
eligible: plan.eligible_count,
mass: plan.mass,
quota,
n_genomes,
fast_mode,
spec,
excluded,
&mut on_layer,
)?;
resolved,
kept,
});
pb.inc(1);
}
pb.finish_and_clear();
Ok(total_kept)
Ok(report)
}
/// One layer's share of [`sample_index`] — see the module docs for the
/// One layer's share of [`sample_index`], as `(families drawn and resolved,
/// families kept after masking)` — see the module docs for the
/// per-family accept/reject/mark rules and the round-batching scheme. Each
/// eligible family is accepted with probability
/// `quota / mass × pair weight × entropy kernel` (capped at 1): the pair
@@ -303,11 +369,11 @@ fn sample_layer(
spec: SamplingSpec,
excluded: &[bool],
on_layer: &mut impl FnMut(usize, usize, usize, Arc<Vec<SurvivingFamily>>),
) -> OKIResult<usize> {
) -> OKIResult<(usize, usize)> {
let LayerPlan { partition, layer: layer_idx, n_minorants, mass, .. } = *plan;
let eligible = &mut plan.eligible;
if n_minorants == 0 || mass <= 0.0 {
return Ok(0);
return Ok((0, 0));
}
let p0 = quota as f64 / mass;
let layer_dir = cache
@@ -435,5 +501,5 @@ fn sample_layer(
on_layer(partition, layer_idx, raw_index, Arc::new(survivors));
}
Ok(kept)
Ok((resolved, kept))
}
@@ -12,10 +12,13 @@ use obikindex::{OKIError, OKIResult};
use obisys::progress_bar;
use crate::siblings::algorithms::{
AnnexRequest, ensure_layer_family_annexes, ensure_layer_pair_weight_annexes,
inverse_overlap_matrix, is_fast_mode,
AnnexRequest, build_layer_sibling_annex, ensure_layer_family_annexes,
ensure_layer_pair_weight_annexes, inverse_overlap_matrix, is_fast_mode,
};
use crate::siblings::{
ANNEX_FILE_NAME, FAMILY_OVERLAP_FILE_NAMES, FamilyOverlap, FamilyOverlapAccumulator,
QUARTET_ANNEX_FILE_NAMES, ScalarKind, family_overlap_file_name,
};
use crate::siblings::{FamilyOverlap, FamilyOverlapAccumulator, family_overlap_file_name};
pub(crate) trait SiblingBuilder {
/// Build, for every cached layer, the annexes `request` asks for that
@@ -26,13 +29,129 @@ pub(crate) trait SiblingBuilder {
/// a missing entropy, quartet or overlap annex triggers one scan that
/// feeds every missing one, per `ensure_layer_family_annexes`'s own docs,
/// then a second scan builds the missing pair weights, which need the
/// finished overlap matrix. A no-op, cheap existence-check pass, once
/// everything requested already exists.
/// finished overlap matrix. Makes sure the sibling annex exists first
/// ([`ensure_sibling_annex`](Self::ensure_sibling_annex)). A no-op, cheap
/// existence-check pass that takes no lock, once everything requested
/// already exists; otherwise the build runs under the index's exclusive
/// lock ([`KmerIndex::with_lock`](obikindex::KmerIndex::with_lock)).
fn ensure_family_annexes(&self, request: &AnnexRequest) -> OKIResult<()>;
/// [`ensure_family_annexes`](Self::ensure_family_annexes)'s build step,
/// without the existence pre-check and without the lock: the caller
/// holds it.
fn build_family_annexes(&self, request: &AnnexRequest) -> OKIResult<()>;
/// Build the sibling annex (`siblings.psib`, every annex above is derived
/// from it) if any cached layer lacks it — the state of an index right
/// after `migrate-mphf`, which sets every annex aside. The build writes
/// into every layer, so it runs under the index's exclusive lock
/// ([`KmerIndex::with_lock`](obikindex::KmerIndex::with_lock), a no-op
/// when the caller already holds it); a process that waited for it finds
/// the annex built by the one that held it, and does nothing. A no-op,
/// cheap existence-check pass that takes no lock, once every layer has
/// its annex.
fn ensure_sibling_annex(&self) -> OKIResult<()>;
/// (Re)build the sibling annex of every cached layer, unconditionally,
/// removing the annexes derived from it (entropy, quartet, pair weight,
/// family overlap), which its renumbering makes stale. Takes no lock:
/// the caller holds it ([`SiblingExt::build_sibling_annex`], or
/// [`ensure_sibling_annex`](Self::ensure_sibling_annex)).
fn build_base_sibling_annex(&self) -> OKIResult<()>;
}
impl SiblingBuilder for IndexCache {
fn ensure_sibling_annex(&self) -> OKIResult<()> {
let missing = || self.iter_indexed().any(|(_, _, layer)| !layer.dir().join(ANNEX_FILE_NAME).exists());
if !missing() {
return Ok(());
}
self.index().with_lock(|| {
if !missing() {
return Ok(()); // built by the process we waited for
}
tracing::info!("sibling annex missing — building it");
self.build_base_sibling_annex()
})
}
fn build_base_sibling_annex(&self) -> OKIResult<()> {
let n_parts = self.index().n_partitions();
let fast_mode = is_fast_mode(self);
if !fast_mode {
tracing::warn!(
"index has more than 7 layers per partition — sibling-annex fast layer lookup \
disabled; consider compacting this index"
);
}
let pb = progress_bar("sibling_annex", self.n_partition() as u64, "partitions");
let mut total_slots: u64 = 0;
for part in self.partitions() {
let n_layer = self.n_layer(part).unwrap_or(0);
let mut part_slots: u64 = 0;
for l in 0..n_layer {
let layer = self
.get_layer(part, l)
.expect("cache-consistent: n_layer(part) already bounds l");
// Invalidate this layer's entropy, quartet and pair-weight
// annexes, if any — they were built against the sibling
// annex's *previous* `family_idx` numbering
// (`non_monomorphic_reindex_layer` reads the sibling
// annex), and are stale once the new one exists.
// `ensure_family_annexes` only builds what's *missing*, so
// the old files must be removed here for them to be rebuilt
// rather than silently left stale. Removed *before* the new
// annex is built and published (atomically, at its end): a
// build killed in between leaves the previous annex with no
// derived files — still consistent, rebuilt on demand —
// never a new annex next to derived files computed against
// the old numbering.
for name in ScalarKind::Entropy
.file_names()
.into_iter()
.chain(ScalarKind::PairWeight.file_names())
.chain(QUARTET_ANNEX_FILE_NAMES)
{
let derived_path = layer.dir().join(name);
if derived_path.exists() {
std::fs::remove_file(&derived_path).map_err(OKIError::Io)?;
}
}
part_slots += build_layer_sibling_annex(self, &layer, l, n_parts, fast_mode)?;
}
total_slots += part_slots;
pb.inc(1);
pb.set_message(format!(
"partition {part}: {part_slots} kmers ({total_slots} total)"
));
}
pb.finish_and_clear();
tracing::info!("sibling annex built — {total_slots} kmers across {n_parts} partitions");
// The family-overlap annexes are index-level, not per-layer (see
// their own module docs) — any layer's sibling annex changing makes
// them all stale (and, with them, every pair weight computed from
// them, removed per layer above), so they are invalidated once here
// rather than per-layer above.
for name in FAMILY_OVERLAP_FILE_NAMES {
let overlap_path = self.index().dir().join(name);
if overlap_path.exists() {
std::fs::remove_file(&overlap_path).map_err(OKIError::Io)?;
}
}
Ok(())
}
fn ensure_family_annexes(&self, request: &AnnexRequest) -> OKIResult<()> {
self.ensure_sibling_annex()?;
if request.is_satisfied(self) {
return Ok(());
}
self.index().with_lock(|| self.build_family_annexes(request))
}
fn build_family_annexes(&self, request: &AnnexRequest) -> OKIResult<()> {
let n_genomes = self.meta().genomes().len();
let fast_mode = is_fast_mode(self);
let overlap_path = |no_ambiguity: bool| self.index().dir().join(family_overlap_file_name(no_ambiguity));
@@ -11,18 +11,12 @@ use std::path::Path;
use obikidxcache::index_cache::IndexCache;
use obikindex::{OKIError, OKIResult};
use obisys::progress_bar;
use crate::siblings::algorithms::{
AnnexRequest, Selection, SiblingAnnexStats, build_layer_sibling_annex, family_entropy,
family_entropy_4, is_fast_mode, non_monomorphic_reindex_layer, scan_layer_families,
sibling_annex_stats, sibling_family_size_histogram,
AnnexRequest, EntropySummary, SiblingAnnexStats, scan_family_entropies, sibling_annex_stats, sibling_family_size_histogram,
};
use crate::siblings::extensions::SiblingBuilder;
use crate::siblings::{
FAMILY_OVERLAP_FILE_NAMES, FamilyOverlap, QUARTET_ANNEX_FILE_NAMES, ScalarKind,
family_overlap_file_name,
};
use crate::siblings::{FamilyOverlap, family_overlap_file_name};
pub trait SiblingExt {
/// Build the sibling-count/minorant annex for every layer of every
@@ -66,6 +60,18 @@ pub trait SiblingExt {
/// `--entropy-bias` yet (see the project memory on this).
fn shannon_entropy_csv(&self, path: &Path) -> OKIResult<()>;
/// The same scan as [`shannon_entropy_csv`](Self::shannon_entropy_csv), but
/// written as a distribution instead of one row per family: a CSV
/// (`entropy,n_genomes_present,n_genomes_unambiguous,bin_start,n_families`)
/// counting the families per `entropy15` (all carriers) or `entropy4`
/// (single-member genomes; families with none are left out of its lines),
/// number of genomes present, number of those without ambiguity, and
/// 0.05-bit entropy bin. At most a few hundred thousand lines whatever the
/// size of the index — enough to answer how entropy varies with the number
/// of carriers, or how many carriers `--no-ambiguity` loses, without the
/// tens of gigabytes the per-family file takes.
fn shannon_entropy_summary(&self, path: &Path) -> OKIResult<()>;
/// The global family-size histogram alone (`SiblingAnnexStats::counts`,
/// no `per_genome`) — reads only the already-built annex
/// (`FamilyMask::siblings()` + `is_minorant()`, already-stored bits,
@@ -100,147 +106,50 @@ pub trait SiblingExt {
impl SiblingExt for IndexCache {
fn build_sibling_annex(&self) -> OKIResult<()> {
let n_parts = self.index().n_partitions();
let fast_mode = is_fast_mode(self);
if !fast_mode {
tracing::warn!(
"index has more than 7 layers per partition — sibling-annex fast layer lookup \
disabled; consider compacting this index"
);
}
let pb = progress_bar("sibling_annex", self.n_partition() as u64, "partitions");
let mut total_slots: u64 = 0;
for part in self.partitions() {
let n_layer = self.n_layer(part).unwrap_or(0);
let mut part_slots: u64 = 0;
for l in 0..n_layer {
let layer = self
.get_layer(part, l)
.expect("cache-consistent: n_layer(part) already bounds l");
part_slots += build_layer_sibling_annex(self, &layer, l, n_parts, fast_mode)?;
// Invalidate this layer's entropy and quartet annexes, if
// any — they were built against the sibling annex's
// *previous* `family_idx` numbering
// (`non_monomorphic_reindex_layer` reads the sibling
// annex), now stale. `ensure_family_annexes` below only
// builds what's *missing*, so the old files must be removed
// here for them to be rebuilt rather than silently left
// stale.
for name in ScalarKind::Entropy
.file_names()
.into_iter()
.chain(ScalarKind::PairWeight.file_names())
.chain(QUARTET_ANNEX_FILE_NAMES)
{
let entropy_path = layer.dir().join(name);
if entropy_path.exists() {
std::fs::remove_file(&entropy_path).map_err(OKIError::Io)?;
}
}
}
total_slots += part_slots;
pb.inc(1);
pb.set_message(format!(
"partition {part}: {part_slots} kmers ({total_slots} total)"
));
}
pb.finish_and_clear();
tracing::info!("sibling annex built — {total_slots} kmers across {n_parts} partitions");
// The family-overlap annexes are index-level, not per-layer (see
// their own module docs) — any layer's sibling annex changing makes
// them all stale (and, with them, every pair weight computed from
// them, removed per layer above), so they are invalidated once here
// rather than per-layer above.
for name in FAMILY_OVERLAP_FILE_NAMES {
let overlap_path = self.index().dir().join(name);
if overlap_path.exists() {
std::fs::remove_file(&overlap_path).map_err(OKIError::Io)?;
}
}
// Precompute the entropy and family-overlap annexes too, now that
// every layer's fresh sibling annex is on disk
// (`non_monomorphic_reindex_layer`, which both depend on, reads
// it) — same routine-rebuild footing as the sibling annex itself,
// rather than leaving them to be built lazily on first
// `--shannon`/`--entropy-bias`/`--min-shared-family`/
// `--family-overlap` use.
tracing::info!("building entropy and quartet (both --no-ambiguity conventions) and family-overlap annexes");
self.ensure_family_annexes(&AnnexRequest::everything())?;
Ok(())
self.index().with_lock(|| {
self.build_base_sibling_annex()?;
// Precompute the entropy, quartet, pair-weight and family-overlap
// annexes too, now that every layer's fresh sibling annex is on
// disk (`non_monomorphic_reindex_layer`, which they depend on,
// reads it) — same routine-rebuild footing as the sibling annex
// itself, rather than leaving them to be built lazily on first
// `--shannon`/`--entropy`/`--quartet`/`--pair-balance`/
// `--min-shared-family`/`--family-overlap` use.
tracing::info!("building entropy, quartet, pair-weight (both --no-ambiguity conventions) and family-overlap annexes");
self.ensure_family_annexes(&AnnexRequest::everything())
})
}
fn shannon_entropy_csv(&self, path: &Path) -> OKIResult<()> {
// Needs only the sibling annex (`non_monomorphic_reindex_layer`
// below): this method runs its own, differently-scoped scan and
// recomputes every value it writes, so it neither reads nor
// requires the entropy/family-overlap annexes.
let n_genomes = self.meta().genomes().len();
let fast_mode = is_fast_mode(self);
let mut f = BufWriter::new(std::fs::File::create(path).map_err(OKIError::Io)?);
writeln!(f, "partition,layer,family_idx,entropy15,entropy4,family_size,n_genomes_present,n_genomes_unambiguous")
.map_err(OKIError::Io)?;
scan_family_entropies(self, |row| {
writeln!(
f,
"{},{},{},{:.6},{:.6},{},{},{}",
row.partition,
row.layer,
row.family_idx,
row.entropy15,
row.entropy4,
row.family_size,
row.n_present,
row.n_unambiguous
)
})?;
f.flush().map_err(OKIError::Io)
}
let pb = progress_bar("shannon_entropy", self.n_partition() as u64, "partitions");
for part in self.partitions() {
let n_layer = self.n_layer(part).unwrap_or(0);
for l in 0..n_layer {
let layer = self
.get_layer(part, l)
.expect("cache-consistent: n_layer(part) already bounds l");
// Bounded to non-monomorphic minorants only, same reasoning
// as `ensure_layer_entropy_annex`'s own docs: monomorphism
// is already known from the annex bits alone (~98% of a
// layer's minorants, measured on real data), so there is no
// reason to pay `scan_layer_families`'s expensive
// cross-partition resolution for families this method
// skips right below anyway.
let reindex = non_monomorphic_reindex_layer(layer.dir())?;
let eligible: std::collections::HashSet<usize> = reindex.keys().copied().collect();
let mut write_err = None;
scan_layer_families(
self,
part,
l,
n_genomes,
fast_mode,
&Selection::Some(&eligible),
|family_idx, mask, genome_mask| {
if write_err.is_some() {
return;
}
debug_assert!(
mask.family_size() >= 2,
"Selection::Some(eligible) must only visit non-monomorphic minorants"
);
let Some((h15, n_present)) = family_entropy(genome_mask) else {
return;
};
let (h4, n_unambiguous) = family_entropy_4(genome_mask).unwrap_or((0.0, 0));
if let Err(e) = writeln!(
f,
"{part},{l},{family_idx},{h15:.6},{h4:.6},{},{n_present},{n_unambiguous}",
mask.family_size()
) {
write_err = Some(e);
}
},
)?;
if let Some(e) = write_err {
return Err(OKIError::Io(e));
}
}
pb.inc(1);
}
pb.finish_and_clear();
Ok(())
fn shannon_entropy_summary(&self, path: &Path) -> OKIResult<()> {
let mut summary = EntropySummary::default();
scan_family_entropies(self, |row| {
summary.add(row);
Ok(())
})?;
let mut f = BufWriter::new(std::fs::File::create(path).map_err(OKIError::Io)?);
summary.write(&mut f).map_err(OKIError::Io)?;
f.flush().map_err(OKIError::Io)
}
fn sibling_family_size_histogram(&self) -> OKIResult<[u64; 4]> {
+3 -1
View File
@@ -23,6 +23,7 @@ use std::io;
use std::path::Path;
use memmap2::Mmap;
use obicompactvec::{building_path, publish_built};
use crate::siblings::helpers::{is_carrier, triangle_index};
@@ -166,7 +167,8 @@ impl FamilyOverlapAccumulator {
for &c in &self.counts {
buf.extend_from_slice(&c.to_le_bytes());
}
std::fs::write(path, buf)
std::fs::write(building_path(path), buf)?;
publish_built(path)
}
}
@@ -20,9 +20,10 @@
use std::fs::{File, OpenOptions};
use std::io;
use std::path::Path;
use std::path::{Path, PathBuf};
use memmap2::{Mmap, MmapMut};
use obicompactvec::{building_path, publish_built};
// Header: magic(4) + flags(4) + n(8) = 16 bytes. Data (4 bytes/entry) follows.
// `flags` bit 0 = `no_ambiguity` (the counting convention the values were
@@ -125,6 +126,7 @@ impl FamilyScalarAnnex {
pub struct FamilyScalarAnnexBuilder {
mmap: MmapMut,
path: PathBuf,
}
impl FamilyScalarAnnexBuilder {
@@ -139,7 +141,7 @@ impl FamilyScalarAnnexBuilder {
.write(true)
.create(true)
.truncate(true)
.open(path)?;
.open(building_path(path))?;
file.set_len(file_size as u64)?;
let mut mmap = unsafe { MmapMut::map_mut(&file)? };
mmap[0..4].copy_from_slice(&kind.magic());
@@ -150,7 +152,7 @@ impl FamilyScalarAnnexBuilder {
let off = HEADER_SIZE + i * 4;
mmap[off..off + 4].copy_from_slice(&SENTINEL.to_le_bytes());
}
Ok(Self { mmap })
Ok(Self { mmap, path: path.to_path_buf() })
}
pub(crate) fn set(&mut self, idx: usize, value: f32) {
@@ -159,8 +161,12 @@ impl FamilyScalarAnnexBuilder {
self.mmap[off..off + 4].copy_from_slice(&value.to_le_bytes());
}
/// Flushes, then publishes the file under its final name: until then it
/// lives under `building_path`, so a killed build never leaves a file
/// `path.exists()` would take for a finished annex.
pub(crate) fn close(self) -> io::Result<()> {
self.mmap.flush()
self.mmap.flush()?;
publish_built(&self.path)
}
}
+2 -2
View File
@@ -42,8 +42,8 @@ pub use siblingannex::{FamilyMask, SiblingAnnex, SiblingAnnexBuilder};
pub use algorithms::{
AlignmentReducer, BasePairTally, CardinalityTally, EntropyBias, GammaShape, LayerReducer,
PairwiseReducer, RawSnpDistanceOutput, SamplingSpec, SankoffBundle, SiblingAnnexStats,
SnpAlignment,
LayerReport, PairwiseReducer, RawSnpDistanceOutput, SamplePlan, SampleStats,
SampleStatsReducer, SamplingSpec, SankoffBundle, SiblingAnnexStats, SnpAlignment,
SnpDistanceKind, cardinality_transition_probs, composition_transition_probs, distance_matrix,
pairwise_cost_matrix, sampled_scan,
};
+26 -3
View File
@@ -53,6 +53,7 @@ use std::path::{Path, PathBuf};
use std::sync::atomic::AtomicU16;
use memmap2::{Mmap, MmapMut};
use obicompactvec::{building_path, publish_built};
use obikseq::CanonicalKmer;
@@ -268,6 +269,21 @@ impl SiblingAnnex {
Ok(Self { mmap, n, path: path.to_path_buf() })
}
/// The annex of the layer directory `layer_dir`. Same as
/// [`open`](Self::open), except that an absent file is reported with its
/// path and the way to build it, rather than a bare `NotFound` — the
/// state an index is in right after `migrate-mphf` set its annexes aside.
pub fn open_in(layer_dir: &Path) -> io::Result<Self> {
let path = layer_dir.join(super::ANNEX_FILE_NAME);
Self::open(&path).map_err(|e| match e.kind() {
io::ErrorKind::NotFound => io::Error::new(
io::ErrorKind::NotFound,
format!("no sibling annex at {} — build it with `phylo --sibling-annex`", path.display()),
),
_ => e,
})
}
pub fn path(&self) -> &Path { &self.path }
pub fn len(&self) -> usize { self.n }
pub fn is_empty(&self) -> bool { self.n == 0 }
@@ -283,6 +299,7 @@ impl SiblingAnnex {
pub struct SiblingAnnexBuilder {
mmap: MmapMut,
path: PathBuf,
}
impl SiblingAnnexBuilder {
@@ -292,14 +309,14 @@ impl SiblingAnnexBuilder {
let file_size = HEADER_SIZE + n * 2;
let file = OpenOptions::new()
.read(true).write(true).create(true).truncate(true)
.open(path)?;
.open(building_path(path))?;
file.set_len(file_size as u64)?;
let mut mmap = unsafe { MmapMut::map_mut(&file)? };
mmap[0..4].copy_from_slice(&MAGIC);
mmap[4..8].copy_from_slice(&[0u8; 4]);
mmap[8..16].copy_from_slice(&(n as u64).to_le_bytes());
// Data region left at 0 by `set_len`/mmap — the sentinel value.
Ok(Self { mmap })
Ok(Self { mmap, path: path.to_path_buf() })
}
/// `None` means the slot has not (yet) been computed — see module
@@ -336,7 +353,13 @@ impl SiblingAnnexBuilder {
self.mmap[off..off + 2].copy_from_slice(&mask.encode().to_le_bytes());
}
pub fn close(self) -> io::Result<()> { self.mmap.flush() }
/// Flushes, then publishes the annex under its final name — until then
/// it lives under `building_path`, so a killed build leaves no
/// `siblings.psib` for a reader to trust.
pub fn close(self) -> io::Result<()> {
self.mmap.flush()?;
publish_built(&self.path)
}
}
#[cfg(test)]
+17 -11
View File
@@ -1,13 +1,12 @@
use std::collections::HashMap;
use obikseq::CanonicalKmer;
use obiread::record::SeqRecord;
use obiskbuilder::SuperKmerIter;
use crate::partition_kmers::PartitionKmers;
use crate::query_layer::KmerDesc;
/// A batch of query sequences, with k-mers deduplicated directly (not just at
/// the superkmer level) and pre-split by partition.
/// A batch of query sequences, with k-mers grouped by canonical k-mer (not just
/// at the superkmer level) and pre-split by partition.
///
/// Superkmer *construction* (`SuperKmerIter`) is still required — it's the
/// mechanism that computes minimizers and partition routing — but the dedup
@@ -23,8 +22,8 @@ pub(crate) struct QueryBatch {
pub(crate) seqs: Vec<Vec<u8>>,
/// Total kmer count per sequence (used for `--detail` coverage allocation).
pub(crate) n_kmers: Vec<u32>,
/// Deduplicated k-mer occurrences, one map per partition.
pub(crate) by_partition: Vec<HashMap<CanonicalKmer, Vec<KmerDesc>>>,
/// K-mer occurrences grouped by k-mer, one entry per partition.
pub(crate) by_partition: Vec<PartitionKmers>,
}
impl QueryBatch {
@@ -41,20 +40,20 @@ impl QueryBatch {
let mut seqs = Vec::with_capacity(records.len());
let mut n_kmers = Vec::with_capacity(records.len());
let mask = (n_partitions as u64) - 1;
let mut by_partition: Vec<HashMap<CanonicalKmer, Vec<KmerDesc>>> =
(0..n_partitions).map(|_| HashMap::new()).collect();
let mut occurrences: Vec<Vec<(CanonicalKmer, KmerDesc)>> =
(0..n_partitions).map(|_| Vec::new()).collect();
for (seq_idx, record) in records.into_iter().enumerate() {
let mut kmer_offset = 0u32;
for rsk in SuperKmerIter::new(&record.normalized, k, level_max, theta) {
let part_idx = (rsk.minimizer().seq_hash() & mask) as usize;
let map = &mut by_partition[part_idx];
let part = &mut occurrences[part_idx];
for (j, kmer) in rsk.superkmer().iter_canonical_kmers().enumerate() {
map.entry(kmer).or_default().push(KmerDesc {
part.push((kmer, KmerDesc {
seq_idx: seq_idx as u32,
pos: kmer_offset + j as u32,
});
}));
}
let n = (rsk.seql() - k + 1) as u32;
kmer_offset += n;
@@ -65,6 +64,13 @@ impl QueryBatch {
n_kmers.push(kmer_offset);
}
// One partition at a time: each 16-byte-per-occurrence buffer is freed
// as soon as its compact form exists.
let by_partition = occurrences
.into_iter()
.map(PartitionKmers::from_occurrences)
.collect();
Self {
ids,
seqs,
+6 -31
View File
@@ -3,7 +3,6 @@ use std::time::Instant;
use obikidxcache::index_cache::IndexCache;
use obikindex::GenomeInfo;
use obikrope::Rope;
use obikseq::CanonicalKmer;
use obiread::record::parse_chunk;
use tracing::debug;
@@ -62,37 +61,13 @@ pub fn process_chunk(
let batch = QueryBatch::from_records(records, k, 6, 0.7, n_partitions);
let n_seqs = batch.ids.len();
// Estimate QueryBatch::by_partition's actual memory footprint: the
// k-mer-level dedup map — one HashMap<CanonicalKmer, Vec<KmerDesc>> per
// partition, sized by *unique* k-mers, not shrunk by dedup. On real
// workloads with a low intra-chunk duplication rate this can dwarf every
// other per-chunk structure, including the sparse Findere ones logged
// further down. Measured by allocated capacity, not logical length, to
// reflect real memory pressure (HashMap/Vec growth slack) — `by_partition`
// is alive for the entire process_chunk call (never drained, only
// iterated by reference), so this is its footprint for the whole chunk
// lifetime, not a transient.
let hashmap_slot_bytes = (std::mem::size_of::<CanonicalKmer>()
+ std::mem::size_of::<Vec<crate::query_layer::KmerDesc>>()
+ 1) as u64; // +1 ≈ hashbrown control byte per slot
let by_partition_map_bytes: u64 = batch
.by_partition
.iter()
.map(|m| m.capacity() as u64 * hashmap_slot_bytes)
.sum();
let by_partition_desc_bytes: u64 = batch
.by_partition
.iter()
.flat_map(|m| m.values())
.map(|v| v.capacity() as u64 * std::mem::size_of::<crate::query_layer::KmerDesc>() as u64)
.sum();
let by_partition_bytes = by_partition_map_bytes + by_partition_desc_bytes;
// `by_partition` is alive for the entire process_chunk call (never
// drained, only iterated by reference): this is its footprint for the
// whole chunk lifetime. Measured by allocated capacity, not logical length.
debug!(
n_unique_kmers_total = batch.by_partition.iter().map(|m| m.len() as u64).sum::<u64>(),
by_partition_map_bytes,
by_partition_desc_bytes,
by_partition_bytes,
n_unique_kmers_total = batch.by_partition.iter().map(|p| p.len() as u64).sum::<u64>(),
n_occurrences_total = batch.by_partition.iter().map(|p| p.n_occurrences() as u64).sum::<u64>(),
by_partition_bytes = batch.by_partition.iter().map(|p| p.allocated_bytes() as u64).sum::<u64>(),
chunk_bytes,
"by_partition memory retained"
);
+2
View File
@@ -5,6 +5,7 @@
//! the data model, never the reverse), same pattern as `obikindexer` for the
//! build side.
mod partition_kmers;
mod query_layer;
mod batch;
mod chunk;
@@ -12,5 +13,6 @@ mod findere;
mod smer_index;
mod output;
pub use partition_kmers::PartitionKmers;
pub use query_layer::{KmerDesc, QueryHit, QueryPartition, QueryStats};
pub use chunk::process_chunk;
+82
View File
@@ -0,0 +1,82 @@
//! The k-mers of one query chunk routed to one partition, grouped by k-mer.
//!
//! Replaces a `HashMap<CanonicalKmer, Vec<KmerDesc>>`: with genomic reads
//! almost every k-mer occurs once, so the map paid a table slot, a growth
//! slack and a separate heap `Vec` for each unique k-mer (~70 bytes measured)
//! for a dedup that removes under 10 % of the occurrences. Here the
//! occurrences are sorted by k-mer and stored as three flat arrays — one
//! 8-byte k-mer and one 4-byte offset per unique k-mer, one 8-byte
//! [`KmerDesc`] per occurrence — about 20 bytes per unique k-mer, no
//! per-k-mer allocation.
use obikseq::CanonicalKmer;
use crate::query_layer::KmerDesc;
/// Unique k-mers of one partition (sorted by raw value) and, for each, the
/// slice of its occurrences (sorted by `(seq_idx, pos)`, i.e. query order).
pub struct PartitionKmers {
kmers: Vec<CanonicalKmer>,
/// `descs[starts[i]..starts[i + 1]]` are the occurrences of `kmers[i]`.
starts: Vec<u32>,
descs: Vec<KmerDesc>,
}
impl PartitionKmers {
pub fn empty() -> Self {
Self { kmers: Vec::new(), starts: vec![0], descs: Vec::new() }
}
/// Group `occurrences` by k-mer. Consumes the (16 bytes per occurrence)
/// input; after the call only the compact arrays remain.
pub fn from_occurrences(mut occurrences: Vec<(CanonicalKmer, KmerDesc)>) -> Self {
occurrences.sort_unstable_by_key(|(kmer, d)| (kmer.raw(), d.seq_idx, d.pos));
let mut kmers = Vec::new();
let mut starts = Vec::new();
let mut descs = Vec::with_capacity(occurrences.len());
for (i, (kmer, desc)) in occurrences.iter().enumerate() {
if kmers.last() != Some(kmer) {
kmers.push(*kmer);
starts.push(offset(i));
}
descs.push(*desc);
}
starts.push(offset(descs.len()));
kmers.shrink_to_fit();
starts.shrink_to_fit();
Self { kmers, starts, descs }
}
/// Number of unique k-mers.
pub fn len(&self) -> usize {
self.kmers.len()
}
pub fn is_empty(&self) -> bool {
self.kmers.is_empty()
}
/// Total number of occurrences (before grouping).
pub fn n_occurrences(&self) -> usize {
self.descs.len()
}
/// `(kmer, its occurrences)` for every unique k-mer, in k-mer order.
pub fn iter(&self) -> impl Iterator<Item = (CanonicalKmer, &[KmerDesc])> + '_ {
self.kmers.iter().enumerate().map(|(i, &kmer)| {
(kmer, &self.descs[self.starts[i] as usize..self.starts[i + 1] as usize])
})
}
/// Heap bytes held (by allocated capacity).
pub fn allocated_bytes(&self) -> usize {
self.kmers.capacity() * std::mem::size_of::<CanonicalKmer>()
+ self.starts.capacity() * std::mem::size_of::<u32>()
+ self.descs.capacity() * std::mem::size_of::<KmerDesc>()
}
}
fn offset(i: usize) -> u32 {
u32::try_from(i).expect("more than 2^32 k-mer occurrences in one partition of one chunk — lower --chunk-size")
}
+44 -10
View File
@@ -3,6 +3,8 @@ use std::collections::HashMap;
use obikidxcache::index_cache::IndexCache;
use obikseq::CanonicalKmer;
use crate::partition_kmers::PartitionKmers;
// ── KmerDesc — one occurrence of a k-mer in the query batch ──────────────────
/// Describes one occurrence of a (deduplicated) k-mer in the query batch:
@@ -100,7 +102,7 @@ pub trait QueryPartition {
fn query_partition_with<F>(
&self,
part_idx: usize,
kmers: &HashMap<CanonicalKmer, Vec<KmerDesc>>,
kmers: &PartitionKmers,
n_genomes: usize,
on_event: F,
) -> QueryStats
@@ -112,7 +114,7 @@ impl QueryPartition for IndexCache {
fn query_partition_with<F>(
&self,
part_idx: usize,
kmers: &HashMap<CanonicalKmer, Vec<KmerDesc>>,
kmers: &PartitionKmers,
n_genomes: usize,
mut on_event: F,
) -> QueryStats
@@ -131,19 +133,51 @@ impl QueryPartition for IndexCache {
};
// ── Stage 1: MPHF-only pass, bucket hits by (layer_idx, slot) ────────
let mut by_layer: Vec<HashMap<usize, &Vec<KmerDesc>>> =
let mut by_layer: Vec<HashMap<usize, &[KmerDesc]>> =
(0..n_layer).map(|_| HashMap::new()).collect();
for (kmer, descs) in kmers {
stats.n_unique_kmers += 1;
// Layer-major over fixed-size chunks of the (unordered) kmer map:
// each layer sees the chunk's still-unresolved kmers as one batch, so
// its MPHF lookups are prefetched together; a kmer found in an
// earlier layer never reaches the later ones (first hit wins, as
// before). Chunked, not collected — the map is never copied whole.
const CHUNK: usize = 4096;
let mut entries = kmers.iter();
let mut pending: Vec<(CanonicalKmer, &[KmerDesc])> = Vec::with_capacity(CHUNK);
let mut batch: Vec<CanonicalKmer> = Vec::with_capacity(CHUNK);
let mut found: Vec<Option<usize>> = Vec::with_capacity(CHUNK);
loop {
pending.clear();
pending.extend(entries.by_ref().take(CHUNK));
if pending.is_empty() {
break;
}
stats.n_unique_kmers += pending.len();
for l in 0..n_layer {
stats.n_mphf_calls += 1;
if let Some(slot) = self.find_in_layer(part_idx, l, *kmer) {
by_layer[l].insert(slot, descs);
on_event(QueryHit::Found(descs));
stats.n_hits += 1;
if pending.is_empty() {
break;
}
let layer = self
.get_layer(part_idx, l)
.expect("layer within n_layer(part_idx)");
batch.clear();
batch.extend(pending.iter().map(|&(k, _)| k));
stats.n_mphf_calls += batch.len();
layer.find_slot_batch(&batch, &mut found);
let mut i = 0;
pending.retain(|&(_, descs)| {
let hit = found[i];
i += 1;
match hit {
Some(slot) => {
by_layer[l].insert(slot, descs);
on_event(QueryHit::Found(descs));
stats.n_hits += 1;
false
}
None => true,
}
});
}
}
+19 -5
View File
@@ -87,8 +87,8 @@ fn duplicated_sequence_broadcasts_to_both_seq_indices() {
let batch = batch_from_fasta(&fasta, K, 1);
let shared = batch.by_partition[0]
.values()
.find(|descs| descs.iter().any(|d| d.seq_idx == 0) && descs.iter().any(|d| d.seq_idx == 1));
.iter()
.find(|(_, descs)| descs.iter().any(|d| d.seq_idx == 0) && descs.iter().any(|d| d.seq_idx == 1));
assert!(
shared.is_some(),
@@ -119,9 +119,23 @@ fn partition_routing_is_a_pure_function_of_the_kmer() {
// No k-mer key appears in more than one partition's map.
let mut seen: std::collections::HashSet<CanonicalKmer> = std::collections::HashSet::new();
for map in &batch.by_partition {
for kmer in map.keys() {
assert!(seen.insert(*kmer), "k-mer routed to more than one partition");
for part in &batch.by_partition {
for (kmer, _) in part.iter() {
assert!(seen.insert(kmer), "k-mer routed to more than one partition");
}
}
}
#[test]
fn occurrences_of_a_kmer_come_out_in_query_order() {
// Grouping sorts occurrences; within a k-mer they must stay ordered by
// (seq_idx, pos) — the order the former insertion-ordered `Vec` had, which
// downstream output ordering relied on.
let fasta = format!(">r1\n{SEQ}\n>r2\n{SEQ}\n");
let batch = batch_from_fasta(&fasta, K, 1);
for (_, descs) in batch.by_partition[0].iter() {
assert!(descs.windows(2).all(|w| (w[0].seq_idx, w[0].pos) < (w[1].seq_idx, w[1].pos)));
}
let occ: usize = batch.by_partition[0].iter().map(|(_, d)| d.len()).sum();
assert_eq!(occ as u64, total_occurrences(&batch));
}
+2 -3
View File
@@ -64,12 +64,11 @@ fn query_partition_with_missing_index_dir_returns_default_stats() {
let index = KmerIndex::create(tmp.path().join("idx"), config, None).expect("create index");
let cache = IndexCache::new(Arc::new(index), Some(vec![0]));
let mut kmers: HashMap<CanonicalKmer, Vec<KmerDesc>> = HashMap::new();
// Any well-formed canonical k-mer works here — the call must return
// before ever attempting an MPHF lookup, since the partition has zero
// layers.
let kmer = CanonicalKmer::from_raw_unchecked(0u64);
kmers.insert(kmer, vec![KmerDesc { seq_idx: 0, pos: 0 }]);
let kmers = PartitionKmers::from_occurrences(vec![(kmer, KmerDesc { seq_idx: 0, pos: 0 })]);
let stats = cache.query_partition_with(0, &kmers, 1, |_event| {
panic!("on_event must not be called: no index was built");
@@ -94,7 +93,7 @@ fn query_partition_with_empty_kmers_is_a_noop() {
let index = KmerIndex::create(tmp.path().join("idx"), config, None).expect("create index");
let cache = IndexCache::new(Arc::new(index), Some(vec![0]));
let kmers: HashMap<CanonicalKmer, Vec<KmerDesc>> = HashMap::new();
let kmers = PartitionKmers::empty();
let stats = cache.query_partition_with(0, &kmers, 1, |_event| {
panic!("on_event must not be called on an empty kmer map");
});
+7 -2
View File
@@ -22,8 +22,13 @@ use crate::kmer::{CanonicalKmer, Kmer, KmerError, KLen, KmerLength, KmerOf, MLen
use crate::params::k;
use crate::revcomp_lookup::REVCOMP4;
/// Maximum kmers per stored chunk. Enforces the u8 max-kmer-index field in the binary format.
pub const MAX_KMERS_PER_CHUNK: usize = 256;
/// Maximum kmers per stored chunk.
///
/// 128 = 2^7, not the 256 the u8 length field of the binary format could
/// carry: the exact evidence of a layer (`obikindex::layer::evidence`) stores a
/// kmer's rank inside its unitig chunk on 7 bits, so every rank must be ≤ 127.
/// Longer unitigs are split into consecutive chunks at write time.
pub const MAX_KMERS_PER_CHUNK: usize = 128;
// ── PackedSeq ─────────────────────────────────────────────────────────────────
+4 -4
View File
@@ -70,7 +70,7 @@ fn seql_roundtrip() {
fn binary_test_lengths(k: usize) -> Vec<usize> {
use crate::packed_seq::MAX_KMERS_PER_CHUNK;
// Only single-chunk lengths: seql in [k, MAX_KMERS_PER_CHUNK+k-1].
(k..=k + 5).chain([255, 256, 257, MAX_KMERS_PER_CHUNK + k - 1]).collect()
(k..=k + 5).chain([MAX_KMERS_PER_CHUNK + k - 3, MAX_KMERS_PER_CHUNK + k - 2, MAX_KMERS_PER_CHUNK + k - 1]).collect()
}
#[test]
@@ -94,7 +94,7 @@ fn binary_split_roundtrip() {
use crate::packed_seq::MAX_KMERS_PER_CHUNK;
set_k(4);
let k = crate::params::k();
// seql = MAX_KMERS_PER_CHUNK + k = 260 → n_kmers = 257 > 256 → 2 chunks
// seql = MAX_KMERS_PER_CHUNK + k → n_kmers = MAX + 1 > MAX → 2 chunks
let seql = MAX_KMERS_PER_CHUNK + k;
let mut sk = SuperKmer::from_ascii(&make_seq(seql));
sk.set_count(7);
@@ -108,8 +108,8 @@ fn binary_split_roundtrip() {
assert_eq!(chunk0.count(), 7);
assert_eq!(chunk1.count(), 7);
// Chunks cover the original sequence with k-1 overlap — no kmer lost.
assert_eq!(chunk0.seql(), MAX_KMERS_PER_CHUNK + k - 1); // 259
assert_eq!(chunk1.seql(), k); // 4 (1 kmer)
assert_eq!(chunk0.seql(), MAX_KMERS_PER_CHUNK + k - 1);
assert_eq!(chunk1.seql(), k); // 1 kmer
}
#[test]
+1 -1
View File
@@ -164,7 +164,7 @@ mod tests {
use crate::packed_seq::MAX_KMERS_PER_CHUNK;
set_k(4);
let k = crate::params::k();
let valid_lengths: Vec<usize> = (k..=9).chain([255, 256, 257, MAX_KMERS_PER_CHUNK + k - 1]).collect();
let valid_lengths: Vec<usize> = (k..=9).chain([MAX_KMERS_PER_CHUNK + k - 3, MAX_KMERS_PER_CHUNK + k - 2, MAX_KMERS_PER_CHUNK + k - 1]).collect();
for len in valid_lengths {
let u = Unitig::from_ascii(&make_seq(len));
let mut buf = Vec::new();
+1 -1
View File
@@ -25,7 +25,7 @@ pub use normalize::{normalize_fasta_chunk, normalize_fastq_chunk, normalize_sequ
pub use nucstream::{NucPage, NucPageCursor, open_nuc_stream};
pub use path_iterator::{PathIter, path_iter};
pub use peakreader::PeekReader;
pub use xopen::xopen;
pub use xopen::{estimated_decompressed_len, xopen};
/// Default read block size: 1 MiB.
pub const DEFAULT_BLOCK_SIZE: usize = 1024 * 1024;
+38
View File
@@ -0,0 +1,38 @@
use super::*;
use std::io::Write;
fn temp_file(name: &str, content: &[u8]) -> std::path::PathBuf {
let path = std::env::temp_dir().join(format!("obiread_xopen_{}_{name}", std::process::id()));
std::fs::File::create(&path).unwrap().write_all(content).unwrap();
path
}
#[test]
fn estimate_plain_file_is_its_length() {
let p = temp_file("plain", b">s\nACGT\n");
assert_eq!(estimated_decompressed_len(p.to_str().unwrap()), Some(8));
std::fs::remove_file(p).unwrap();
}
#[test]
fn estimate_scales_compressed_files_by_magic_not_extension() {
// Only the magic bytes matter to the estimate (the content is never decoded).
let mut gz = vec![0x1f, 0x8b, 0x08, 0, 0, 0];
gz.resize(100, 0);
let p = temp_file("looks_plain.txt", &gz);
assert_eq!(estimated_decompressed_len(p.to_str().unwrap()), Some(300));
std::fs::remove_file(p).unwrap();
let mut zst = vec![0x28, 0xb5, 0x2f, 0xfd, 0, 0];
zst.resize(50, 0);
let p = temp_file("zst", &zst);
assert_eq!(estimated_decompressed_len(p.to_str().unwrap()), Some(200));
std::fs::remove_file(p).unwrap();
}
#[test]
fn estimate_is_unknown_for_streams_and_missing_files() {
assert_eq!(estimated_decompressed_len("-"), None);
assert_eq!(estimated_decompressed_len("https://example.org/x.fa.gz"), None);
assert_eq!(estimated_decompressed_len("/nonexistent/obiread/file"), None);
}
+35
View File
@@ -46,6 +46,37 @@ pub fn xopen(source: &str) -> io::Result<MimeTypeGuesser<Box<dyn Read + Send>>>
Ok(MimeTypeGuesser::new(open_raw(source)?))
}
/// Rough size, in bytes, of what [`xopen`] would yield for a local file:
/// its length, multiplied by a typical expansion factor when the magic bytes
/// say it is compressed (gzip ≈ 3×, as measured on FASTQ; bzip2, xz and zstd
/// taken as 4×). Meant for sizing work (chunk sizes), not for allocation —
/// the real ratio varies by a factor of two or more with the content.
///
/// `None` when the size cannot be known without reading the stream: stdin
/// (`-`), an HTTP(S) URL, or a file that cannot be opened.
pub fn estimated_decompressed_len(source: &str) -> Option<u64> {
if source == "-" || source.starts_with("http://") || source.starts_with("https://") {
return None;
}
let expanded = expand_tilde(source);
let mut file = File::open(expanded.as_ref()).ok()?;
let len = file.metadata().ok()?.len();
let mut magic = [0u8; 6];
let n = file.read(&mut magic).ok()?;
let magic = &magic[..n];
let factor = if magic.starts_with(&[0x1f, 0x8b]) {
3
} else if magic.starts_with(b"BZh")
|| magic.starts_with(&[0xfd, b'7', b'z', b'X', b'Z', 0x00])
|| magic.starts_with(&[0x28, 0xb5, 0x2f, 0xfd])
{
4
} else {
1
};
Some(len * factor)
}
// ── internal helpers ──────────────────────────────────────────────────────────
fn http_reader(url: &str) -> io::Result<Box<dyn Read + Send>> {
@@ -71,3 +102,7 @@ fn expand_tilde(path: &str) -> std::borrow::Cow<'_, str> {
}
std::borrow::Cow::Borrowed(path)
}
#[cfg(test)]
#[path = "tests/xopen.rs"]
mod tests;
+4 -2
View File
@@ -28,8 +28,10 @@ impl SKFileMeta {
/// Read the sidecar for `sk_path`. Returns `None` if the sidecar is absent.
pub fn read(sk_path: &Path) -> SKResult<Option<Self>> {
let path = Self::sidecar_path(sk_path);
match std::fs::File::open(&path) {
Ok(f) => Ok(Some(serde_json::from_reader(f).map_err(|e| SKError::InvalidData {
// Read whole, parse from memory: `from_reader` on a bare `File` is one
// syscall per byte.
match std::fs::read(&path) {
Ok(bytes) => Ok(Some(serde_json::from_slice(&bytes).map_err(|e| SKError::InvalidData {
context: "SKFileMeta sidecar",
detail: e.to_string(),
})?)),
+6 -2
View File
@@ -19,8 +19,12 @@ impl SKFileReader {
/// Open a file for reading. Format is auto-detected from magic bytes.
pub fn open<P: AsRef<Path>>(path: P) -> SKResult<Self> {
let path = path.as_ref().to_owned();
let (reader, _fmt) =
niffler::send::get_reader(Box::new(File::open(&path)?))?;
// Buffered: for an uncompressed file niffler hands back the bare
// `File` (after the sniffed bytes), and `read_superkmer` reads each
// field with its own small `read_exact` — one syscall each otherwise.
let (reader, _fmt) = niffler::send::get_reader(Box::new(
std::io::BufReader::with_capacity(1 << 16, File::open(&path)?),
))?;
Ok(Self { path, reader })
}
+3 -1
View File
@@ -104,7 +104,9 @@ fn roundtrip_single() {
fn roundtrip_all_lengths() {
setup();
let bases: Vec<u8> = (0..300).map(|i| b"ACGT"[i % 4]).collect();
for len in (TEST_K..=19).chain([255, 256, 257]) {
// single-chunk lengths only: seql ≤ MAX_KMERS_PER_CHUNK + k - 1
let max_len = obikseq::MAX_KMERS_PER_CHUNK + TEST_K - 1;
for len in (TEST_K..=19).chain([max_len - 2, max_len - 1, max_len]) {
let sk = make_sk(&bases[..len]);
let mut buf = Vec::new();
sk.write_to_binary(&mut buf).unwrap();
+27 -13
View File
@@ -222,31 +222,45 @@ fn iter_indexed_chunk_id_and_rank_two_chunks() {
#[test]
fn short_unitig_not_split() {
// seql=259 → n_kmers=256 = MAX_KMERS_PER_CHUNK → no split
// seql=131 → n_kmers=128 = MAX_KMERS_PER_CHUNK → no split
set_k(4);
let seq: Vec<u8> = (0..259_usize).map(|i| b"ACGT"[i % 4]).collect();
let seq: Vec<u8> = (0..131_usize).map(|i| b"ACGT"[i % 4]).collect();
let (_dir, r) = write_read(&[&seq]);
assert_eq!(r.len(), 1);
assert_eq!(r.seql(0), 259);
assert_eq!(r.seql(0), 131);
}
#[test]
fn long_unitig_split_no_kmer_lost() {
// seql=260 → n_kmers=257 > MAX_KMERS_PER_CHUNK(256) → 2 chunks
// chunk_nucl=259, stride=256
// Chunk 0: nucl 0..259 (259 nucl, 256 kmers)
// Chunk 1: nucl 256..260 (4 nucl, 1 kmer)
// seql=132 → n_kmers=129 > MAX_KMERS_PER_CHUNK(128) → 2 chunks
// chunk_nucl=131, stride=128
// Chunk 0: nucl 0..131 (131 nucl, 128 kmers)
// Chunk 1: nucl 128..132 (4 nucl, 1 kmer)
set_k(4);
let seq: Vec<u8> = (0..260_usize).map(|i| b"ACGT"[i % 4]).collect();
let seq: Vec<u8> = (0..132_usize).map(|i| b"ACGT"[i % 4]).collect();
let (_dir, r) = write_read(&[&seq]);
assert_eq!(r.len(), 2);
assert_eq!(r.seql(0), 259);
assert_eq!(r.seql(0), 131);
assert_eq!(r.seql(1), 4);
// k-1=3 nucleotide overlap → 0 kmers duplicated, 0 kmers lost.
// Last kmer of chunk 0 = original nucl 255..259.
assert!(r.verify_canonical_kmer(0, 255, canonical_of(&seq[255..259])));
// First kmer of chunk 1 = original nucl 256..260 — a different, adjacent kmer.
assert!(r.verify_canonical_kmer(1, 0, canonical_of(&seq[256..260])));
// Last kmer of chunk 0 = original nucl 127..131.
assert!(r.verify_canonical_kmer(0, 127, canonical_of(&seq[127..131])));
// First kmer of chunk 1 = original nucl 128..132 — a different, adjacent kmer.
assert!(r.verify_canonical_kmer(1, 0, canonical_of(&seq[128..132])));
}
#[test]
fn chunk_ranks_fit_the_7_bit_evidence_field() {
// The exact evidence of a layer stores a kmer's rank in its chunk on 7
// bits (obikindex::layer::evidence): a unitig long enough to need several
// chunks must never yield a rank above 127 — a larger one would be
// silently truncated there and the kmer would no longer verify.
set_k(4);
let seq: Vec<u8> = (0..1000_usize).map(|i| b"ACGT"[(i * i + i / 3) % 4]).collect();
let (_dir, r) = write_read(&[&seq]);
assert!(r.len() > 1, "a 997-kmer unitig must be split into several chunks");
let max_rank = r.iter_indexed_canonical_kmers().map(|(_, _, rank)| rank).max().unwrap();
assert!(max_rank <= 127, "rank {max_rank} does not fit 7 bits");
}
// ── block_bits parametrisation ────────────────────────────────────────────────
+50 -37
View File
@@ -24,7 +24,7 @@ use super::{DEFAULT_BLOCK_BITS, MAGIC, idx_path};
/// correct but slower.
pub struct UnitigFileReader {
mmap: Mmap,
block_offsets: Vec<u32>,
block_offsets: BlockOffsets,
n_unitigs: usize,
n_kmers: usize,
k: usize,
@@ -64,7 +64,7 @@ impl UnitigFileReader {
Ok(Self {
mmap,
block_offsets: Vec::new(),
block_offsets: BlockOffsets::none(),
n_unitigs,
n_kmers,
k,
@@ -108,7 +108,7 @@ impl UnitigFileReader {
}
#[inline]
pub fn has_direct_access(&self) -> bool {
!self.block_offsets.is_empty()
!self.block_offsets.is_none()
}
/// Byte offset of record `i` in the mmap.
@@ -119,11 +119,11 @@ impl UnitigFileReader {
fn chunk_start(&self, i: usize) -> usize {
if self.has_direct_access() {
if self.block_bits == 0 {
return self.block_offsets[i] as usize;
return self.block_offsets.get(i) as usize;
}
let block = i >> self.block_bits;
let rem = i & self.mask;
let mut offset = self.block_offsets[block] as usize;
let mut offset = self.block_offsets.get(block) as usize;
for _ in 0..rem {
let seql_minus_k = self.mmap[offset] as usize;
offset += 1 + (seql_minus_k + self.k + 3) / 4;
@@ -292,25 +292,54 @@ impl UnitigFileReader {
}
}
fn read_idx(path: &Path) -> SKResult<(usize, usize, u8, Vec<u32>)> {
let data = std::fs::read(path).map_err(SKError::Io)?;
let mut pos = 0;
/// Block offsets of a `.idx` file, read in place from its memory map: nothing
/// is copied to the heap, and every reader of the same `.idx` shares the same
/// page-cache pages (a copy was ~30 MB per large layer per reader).
struct BlockOffsets {
/// The whole `.idx` file; `None` for a sequential reader (no `.idx`).
mmap: Option<Mmap>,
}
let magic_bytes = data.get(pos..pos + 4).ok_or(SKError::Truncated {
context: "unitig index: magic",
})?;
impl BlockOffsets {
fn none() -> Self {
Self { mmap: None }
}
fn is_none(&self) -> bool {
self.mmap.is_none()
}
/// The `i`-th offset (little-endian `u32` after the fixed header).
/// `read_idx` has checked that the file holds every offset the header
/// announces, so `i` within `0..=n_blocks` cannot run past the map.
#[inline]
fn get(&self, i: usize) -> u32 {
let at = IDX_HEADER_SIZE + 4 * i;
let map = self.mmap.as_ref().expect("direct-access reader has its .idx map");
u32::from_le_bytes(map[at..at + 4].try_into().unwrap())
}
}
/// magic (4) + block_bits (4) + n_unitigs (4) + n_kmers (8).
const IDX_HEADER_SIZE: usize = 20;
fn read_idx(path: &Path) -> SKResult<(usize, usize, u8, BlockOffsets)> {
let file = File::open(path).map_err(SKError::Io)?;
let len = file.metadata().map_err(SKError::Io)?.len() as usize;
if len < IDX_HEADER_SIZE {
return Err(SKError::Truncated { context: "unitig index: header" });
}
let data = unsafe { Mmap::map(&file).map_err(SKError::Io)? };
let magic_bytes = &data[0..4];
if magic_bytes != &MAGIC {
return Err(SKError::BadMagic {
expected: "UIX3",
got: magic_bytes.try_into().unwrap(),
});
}
pos += 4;
let bb_bytes = data.get(pos..pos + 4).ok_or(SKError::Truncated {
context: "unitig index: block_bits",
})?;
let block_bits_u32 = u32::from_le_bytes(bb_bytes.try_into().unwrap());
let block_bits_u32 = u32::from_le_bytes(data[4..8].try_into().unwrap());
if block_bits_u32 > 31 {
return Err(SKError::InvalidData {
context: "unitig index",
@@ -318,31 +347,15 @@ fn read_idx(path: &Path) -> SKResult<(usize, usize, u8, Vec<u32>)> {
});
}
let block_bits = block_bits_u32 as u8;
pos += 4;
let n_bytes = data.get(pos..pos + 4).ok_or(SKError::Truncated {
context: "unitig index: n_unitigs",
})?;
let n_unitigs = u32::from_le_bytes(n_bytes.try_into().unwrap()) as usize;
pos += 4;
let nk_bytes = data.get(pos..pos + 8).ok_or(SKError::Truncated {
context: "unitig index: n_kmers",
})?;
let n_kmers = u64::from_le_bytes(nk_bytes.try_into().unwrap()) as usize;
pos += 8;
let n_unitigs = u32::from_le_bytes(data[8..12].try_into().unwrap()) as usize;
let n_kmers = u64::from_le_bytes(data[12..20].try_into().unwrap()) as usize;
let block_size = 1usize << block_bits;
let n_blocks = (n_unitigs + block_size - 1) >> block_bits;
let n_offsets = n_blocks + 1;
let mut block_offsets = Vec::with_capacity(n_offsets);
for _ in 0..n_offsets {
let off_bytes = data.get(pos..pos + 4).ok_or(SKError::Truncated {
context: "unitig index: block_offsets",
})?;
block_offsets.push(u32::from_le_bytes(off_bytes.try_into().unwrap()));
pos += 4;
if len < IDX_HEADER_SIZE + 4 * n_offsets {
return Err(SKError::Truncated { context: "unitig index: block_offsets" });
}
Ok((n_unitigs, n_kmers, block_bits, block_offsets))
Ok((n_unitigs, n_kmers, block_bits, BlockOffsets { mmap: Some(data) }))
}
+5 -1
View File
@@ -3,6 +3,9 @@
mod budget;
mod lock;
mod lock_slot;
#[cfg(test)]
mod lock_tests;
mod numa;
mod profiler;
mod progress;
@@ -10,7 +13,8 @@ mod resources;
mod stage;
pub use budget::MemoryBudget;
pub use lock::DirLock;
pub use lock::{DirLock, LockMode};
pub use lock_slot::LockSlot;
pub use numa::PartitionRunner;
pub use profiler::start as start_profiler;
pub use progress::{Progress, TracedBar, progress_bar, spinner};
+188 -38
View File
@@ -1,56 +1,206 @@
use std::fs::File;
use std::io;
use std::path::{Path, PathBuf};
use std::time::Duration;
use tracing::info;
/// Exclusive advisory lock on an index directory, held for the duration of
/// any command that writes into an already-existing index (building the
/// sibling annex, merging into a destination, filtering/selecting in place,
/// ...). Two such commands racing on the same directory can otherwise
/// corrupt each other's writes with no error from either side.
/// Advisory readers/writer lock on an index directory, shared between
/// processes (and between independent handles of one process).
///
/// Only the directory actually being *written to* needs a lock — a command
/// like `merge` that reads several source indexes to build one destination
/// only needs to lock the destination.
/// - **Shared** ([`acquire_shared`](DirLock::acquire_shared)): held by
/// anything that reads the index (`query`, `dump`, a `merge` source, a
/// `phylo` run, ...). Any number of shared holders coexist.
/// - **Exclusive** ([`acquire`](DirLock::acquire)): held by anything that
/// writes into an already-existing index (building annexes, `convert`,
/// `pack`, `annotate`, ...). It waits for every shared holder to leave and
/// keeps others out while held. Two writers racing on one directory can
/// otherwise corrupt each other's files with no error from either side,
/// and a reader would see files vanish under it.
///
/// Uses the OS's advisory file lock (`flock` on Unix, `LockFileEx` on
/// Windows) via `std::fs::File::lock`/`try_lock`, not a hand-rolled PID
/// file: the OS releases it automatically on process exit, including a
/// crash — no stale-lock cleanup logic needed.
/// Only the directory actually being *written to* needs an exclusive lock —
/// `merge` holds shared locks on its sources and an exclusive one on its
/// destination.
///
/// # How it is built
///
/// Each mode is one OS advisory file lock (`flock` on Unix, `LockFileEx` on
/// Windows) on `.obikmer.lock`, so a crash releases it with no stale-lock
/// cleanup. On top of that, **every transition** (acquire, release into
/// another mode, promote, demote) happens while holding a *gate* — a second
/// lock file, `.obikmer.gate`, taken for the length of one non-blocking
/// attempt and never while waiting. A blocked request therefore releases the
/// gate, sleeps, and tries again (polling, 20 ms growing to 500 ms), instead
/// of sleeping in the kernel while holding it. Because nobody else can touch
/// the lock state while the gate is held, a shared → exclusive
/// [promotion](DirLock::promote) cannot be overtaken by another writer, which
/// the OS's own `flock` conversion does not guarantee.
///
/// There is no writer preference: a steady stream of readers can delay an
/// exclusive request.
#[derive(Debug)]
pub struct DirLock {
_file: std::fs::File,
file: File,
dir: PathBuf,
mode: LockMode,
}
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum LockMode {
Shared,
Exclusive,
}
const LOCK_FILE: &str = ".obikmer.lock";
const GATE_FILE: &str = ".obikmer.gate";
const CLAIM_FILE: &str = ".obikmer.promote";
const POLL_MIN: Duration = Duration::from_millis(20);
const POLL_MAX: Duration = Duration::from_millis(500);
fn open_file(dir: &Path, name: &str) -> io::Result<File> {
std::fs::create_dir_all(dir)?;
std::fs::OpenOptions::new()
.create(true)
.truncate(false)
.write(true)
.open(dir.join(name))
}
/// The gate: blocks, but its holders only ever run instantaneous,
/// non-blocking operations, so the wait is negligible. Released on drop.
fn gate(dir: &Path) -> io::Result<File> {
let f = open_file(dir, GATE_FILE)?;
f.lock()?;
Ok(f)
}
/// Non-blocking attempt in `mode`; `Ok(false)` means "held by someone else".
fn try_lock(file: &File, mode: LockMode) -> io::Result<bool> {
let r = match mode {
LockMode::Exclusive => file.try_lock(),
LockMode::Shared => file.try_lock_shared(),
};
match r {
Ok(()) => Ok(true),
Err(std::fs::TryLockError::WouldBlock) => Ok(false),
Err(std::fs::TryLockError::Error(e)) => Err(e),
}
}
/// Sleeps between two attempts and logs once, at the first one.
struct Backoff {
delay: Duration,
logged: bool,
}
impl Backoff {
fn new() -> Self {
Self { delay: POLL_MIN, logged: false }
}
fn wait(&mut self, dir: &Path, why: &str) {
if !self.logged {
info!(dir = %dir.display(), "{why}");
self.logged = true;
}
std::thread::sleep(self.delay);
self.delay = (self.delay * 2).min(POLL_MAX);
}
}
impl DirLock {
/// Block until the exclusive lock on `dir` is acquired (creating `dir`
/// and the lock file within it if needed). Logs once if the wait is
/// and the lock files within it if needed). Logs once if the wait is
/// non-trivial, so a blocked command doesn't look silently hung.
pub fn acquire(dir: &std::path::Path) -> std::io::Result<Self> {
let file = Self::open_lock_file(dir)?;
if file.try_lock().is_err() {
info!(dir = %dir.display(), "waiting for another obikmer process to release this index");
file.lock()?;
}
Ok(Self { _file: file })
pub fn acquire(dir: &Path) -> io::Result<Self> {
Self::acquire_mode(dir, LockMode::Exclusive, "waiting for other obikmer processes to release this index")
}
/// Non-blocking probe: `Ok(Some(lock))` if `dir`'s lock was free and is
/// now held by the returned `DirLock`, `Ok(None)` if another process
/// currently holds it (not an error — contention is the expected,
/// common case this exists to distinguish from a genuine I/O failure).
pub fn try_acquire(dir: &std::path::Path) -> std::io::Result<Option<Self>> {
let file = Self::open_lock_file(dir)?;
match file.try_lock() {
Ok(()) => Ok(Some(Self { _file: file })),
Err(_) => Ok(None),
/// Block until a shared lock on `dir` is acquired — until no exclusive
/// holder remains. Logs once if the wait is non-trivial.
pub fn acquire_shared(dir: &Path) -> io::Result<Self> {
Self::acquire_mode(dir, LockMode::Shared, "waiting for an obikmer process writing this index")
}
fn acquire_mode(dir: &Path, mode: LockMode, why: &str) -> io::Result<Self> {
let file = open_file(dir, LOCK_FILE)?;
let mut backoff = Backoff::new();
loop {
{
let _gate = gate(dir)?;
if try_lock(&file, mode)? {
return Ok(Self { file, dir: dir.to_owned(), mode });
}
}
backoff.wait(dir, why);
}
}
fn open_lock_file(dir: &std::path::Path) -> std::io::Result<std::fs::File> {
std::fs::create_dir_all(dir)?;
let lock_path = dir.join(".obikmer.lock");
std::fs::OpenOptions::new()
.create(true)
.truncate(false)
.write(true)
.open(&lock_path)
pub fn mode(&self) -> LockMode {
self.mode
}
/// Non-blocking probe: `Ok(true)` if no exclusive holder exists right
/// now (a shared lock could be taken), `Ok(false)` if one does. Retains
/// nothing either way.
pub fn is_free_of_writer(dir: &Path) -> io::Result<bool> {
let file = open_file(dir, LOCK_FILE)?;
let _gate = gate(dir)?;
try_lock(&file, LockMode::Shared)
}
/// Shared → exclusive. Returns the exclusive lock and whether the
/// promotion was **gapless**: `true` means the lock was never released,
/// so nothing in the directory can have changed since the shared lock
/// was held and whatever the caller read stays valid; `false` means the
/// shared lock had to be given up while waiting, so a writer may have
/// run in between and the caller must re-check what it relied on.
///
/// Gapless promotion is available to one promoter at a time (it holds
/// `.obikmer.promote` while it waits for the other readers to leave).
/// A second concurrent promoter releases its shared lock instead — two
/// promoters each keeping theirs while waiting for the other's would
/// wait for ever. An exclusive lock is returned unchanged.
pub fn promote(self) -> io::Result<(Self, bool)> {
if self.mode == LockMode::Exclusive {
return Ok((self, true));
}
let claim = open_file(&self.dir, CLAIM_FILE)?;
if !try_lock(&claim, LockMode::Exclusive)? {
let dir = self.dir.clone();
drop(self); // gives the shared lock up
return Ok((Self::acquire(&dir)?, false));
}
let mut backoff = Backoff::new();
loop {
{
let _gate = gate(&self.dir)?;
self.file.unlock()?;
if try_lock(&self.file, LockMode::Exclusive)? {
return Ok((Self { mode: LockMode::Exclusive, ..self }, true));
}
// Other readers remain. Nobody can hold the lock exclusively
// without the gate, so taking our shared lock back succeeds.
if !try_lock(&self.file, LockMode::Shared)? {
return Err(io::Error::other("lost the shared index lock during promotion"));
}
}
backoff.wait(&self.dir, "waiting for other readers of this index before writing to it");
}
}
/// Exclusive → shared, without ever leaving the lock free in between.
/// A shared lock is returned unchanged.
pub fn demote(self) -> io::Result<Self> {
if self.mode == LockMode::Shared {
return Ok(self);
}
let held_gate = gate(&self.dir)?;
self.file.unlock()?;
if !try_lock(&self.file, LockMode::Shared)? {
return Err(io::Error::other("could not take the shared index lock back after writing"));
}
drop(held_gate);
Ok(Self { mode: LockMode::Shared, ..self })
}
}
+96
View File
@@ -0,0 +1,96 @@
use std::io;
use std::path::{Path, PathBuf};
use std::sync::Mutex;
use crate::lock::{DirLock, LockMode};
/// The directory lock of one handle on an index: nothing, shared or
/// exclusive, and the transitions between them that a handle needs. What
/// `obikindex::KmerIndex` delegates its locking to, kept apart so that it can
/// be tested without building an index.
///
/// Interior mutability (`&self` methods): a handle is shared behind `Arc`.
/// Not for concurrent use from several threads of one handle — the first
/// section to finish would restore the lock under the others.
#[derive(Debug)]
pub struct LockSlot {
dir: PathBuf,
held: Mutex<Option<DirLock>>,
}
impl LockSlot {
/// A slot holding nothing (a handle that took no lock, or one that
/// reopens a directory its own builder already holds).
pub fn empty(dir: &Path) -> Self {
Self { dir: dir.to_owned(), held: Mutex::new(None) }
}
/// A slot holding a shared lock on `dir`; blocks while a writer is at work.
pub fn shared(dir: &Path) -> io::Result<Self> {
Ok(Self { dir: dir.to_owned(), held: Mutex::new(Some(DirLock::acquire_shared(dir)?)) })
}
/// A slot holding the exclusive lock on `dir`; blocks while anyone else holds it.
pub fn exclusive(dir: &Path) -> io::Result<Self> {
Ok(Self { dir: dir.to_owned(), held: Mutex::new(Some(DirLock::acquire(dir)?)) })
}
/// Whether the slot holds the exclusive lock (a shared one is not write access).
pub fn holds_write_lock(&self) -> bool {
matches!(&*self.held.lock().unwrap(), Some(l) if l.mode() == LockMode::Exclusive)
}
/// Hold the exclusive lock from now on: a no-op if already held that way,
/// a promotion of a shared lock, a plain acquisition of an empty slot.
pub fn acquire_exclusive(&self) -> io::Result<()> {
let held = self.held.lock().unwrap().take();
let lock = match held {
Some(l) => l.promote()?.0,
None => DirLock::acquire(&self.dir)?,
};
*self.held.lock().unwrap() = Some(lock);
Ok(())
}
/// Run `f` with the exclusive lock held:
/// - already exclusive: `f` runs as is (so sections nest);
/// - shared: promoted for the duration of `f`, then demoted back, so the
/// slot keeps the shared lock it came with. The promotion is gapless
/// when no other process promotes at the same time; otherwise other
/// writers may run between the shared lock being given up and the
/// exclusive one being taken. Either way **`f` must re-check, before
/// building, what it is about to build**, and the caller must not hold
/// anything read earlier that a writer could have replaced;
/// - empty: the lock is taken for `f` only.
///
/// Restored also if `f` fails; an I/O error taking or restoring the lock
/// is converted to `E`.
pub fn with_exclusive<T, E: From<io::Error>>(&self, f: impl FnOnce() -> Result<T, E>) -> Result<T, E> {
enum Before {
Exclusive,
Shared,
Nothing,
}
let held = self.held.lock().unwrap().take();
let (lock, before) = match held {
Some(l) if l.mode() == LockMode::Exclusive => (l, Before::Exclusive),
Some(l) => (l.promote()?.0, Before::Shared),
None => (DirLock::acquire(&self.dir)?, Before::Nothing),
};
*self.held.lock().unwrap() = Some(lock);
let out = f();
let lock = self.held.lock().unwrap().take();
match (before, lock) {
(Before::Exclusive, lock) => *self.held.lock().unwrap() = lock,
(Before::Shared, Some(l)) => *self.held.lock().unwrap() = Some(l.demote()?),
_ => {} // taken for `f` only: released here
}
out
}
/// Non-blocking probe: `true` if no process holds the directory
/// exclusively right now (this slot included), `false` if one does.
pub fn is_free_of_writer(&self) -> bool {
DirLock::is_free_of_writer(&self.dir).unwrap_or(false)
}
}
+208
View File
@@ -0,0 +1,208 @@
//! The directory lock is a readers/writer lock: `shared` slots coexist, an
//! `exclusive` one waits for them all and keeps new ones out, and a shared
//! slot can be promoted for a section (`with_exclusive`) and demoted back.
//! No index is involved, and therefore none of the process-wide k-mer
//! parameters other crates' tests share.
use std::io;
use std::path::PathBuf;
use std::sync::atomic::{AtomicUsize, Ordering};
use std::sync::mpsc;
use std::sync::{Arc, Barrier};
use std::thread;
use std::time::Duration;
use crate::lock::{DirLock, LockMode};
use crate::lock_slot::LockSlot;
const SETTLE: Duration = Duration::from_millis(400);
const PATIENCE: Duration = Duration::from_secs(10);
/// A scratch directory removed on drop (no `tempfile` dependency here).
struct Scratch(PathBuf);
impl Scratch {
fn new() -> Self {
static N: AtomicUsize = AtomicUsize::new(0);
let dir = std::env::temp_dir().join(format!(
"obisys-lock-{}-{}",
std::process::id(),
N.fetch_add(1, Ordering::Relaxed)
));
std::fs::create_dir_all(&dir).unwrap();
Self(dir)
}
}
impl Drop for Scratch {
fn drop(&mut self) {
let _ = std::fs::remove_dir_all(&self.0);
}
}
#[test]
fn readers_coexist() {
let d = Scratch::new();
let a = LockSlot::shared(&d.0).unwrap();
let b = LockSlot::shared(&d.0).unwrap();
assert!(!a.holds_write_lock() && !b.holds_write_lock());
assert!(a.is_free_of_writer(), "readers are not writers");
}
#[test]
fn exclusive_is_exclusive_and_released_on_drop() {
let d = Scratch::new();
let w = LockSlot::exclusive(&d.0).unwrap();
assert!(w.holds_write_lock());
assert!(!w.is_free_of_writer());
drop(w);
assert!(LockSlot::empty(&d.0).is_free_of_writer());
}
#[test]
fn a_writer_waits_for_the_readers() {
let d = Scratch::new();
let reader = LockSlot::shared(&d.0).unwrap();
let (tx, rx) = mpsc::channel();
let dir = d.0.clone();
let writer = thread::spawn(move || {
drop(LockSlot::exclusive(&dir).unwrap());
tx.send(()).unwrap();
});
assert!(rx.recv_timeout(SETTLE).is_err(), "the writer must wait for the reader");
drop(reader);
rx.recv_timeout(PATIENCE).expect("the writer proceeds once the reader is gone");
writer.join().unwrap();
}
#[test]
fn a_reader_waits_for_the_writer() {
let d = Scratch::new();
let writer = LockSlot::exclusive(&d.0).unwrap();
let (tx, rx) = mpsc::channel();
let dir = d.0.clone();
let reader = thread::spawn(move || {
drop(LockSlot::shared(&dir).unwrap());
tx.send(()).unwrap();
});
assert!(rx.recv_timeout(SETTLE).is_err(), "the reader must wait for the writer");
drop(writer);
rx.recv_timeout(PATIENCE).expect("the reader proceeds once the writer is gone");
reader.join().unwrap();
}
#[test]
fn with_exclusive_on_a_reader_promotes_then_restores_the_shared_lock() {
let d = Scratch::new();
let slot = LockSlot::shared(&d.0).unwrap();
let out: io::Result<i32> = slot.with_exclusive(|| {
assert!(slot.holds_write_lock());
assert!(!slot.is_free_of_writer());
Ok(7)
});
assert_eq!(out.unwrap(), 7);
assert!(!slot.holds_write_lock());
assert!(slot.is_free_of_writer());
// Still a reader: a writer elsewhere waits for it.
let (tx, rx) = mpsc::channel();
let dir = d.0.clone();
let writer = thread::spawn(move || {
drop(LockSlot::exclusive(&dir).unwrap());
tx.send(()).unwrap();
});
assert!(rx.recv_timeout(SETTLE).is_err(), "the slot must be a reader again");
drop(slot);
rx.recv_timeout(PATIENCE).unwrap();
writer.join().unwrap();
}
#[test]
fn with_exclusive_on_an_empty_slot_holds_the_lock_for_the_section_only() {
let d = Scratch::new();
let slot = LockSlot::empty(&d.0);
slot.with_exclusive::<(), io::Error>(|| {
assert!(slot.holds_write_lock());
Ok(())
})
.unwrap();
assert!(!slot.holds_write_lock());
assert!(slot.is_free_of_writer());
}
#[test]
fn with_exclusive_restores_on_error() {
let d = Scratch::new();
let slot = LockSlot::shared(&d.0).unwrap();
let r: io::Result<()> = slot.with_exclusive(|| Err(io::Error::other("boom")));
assert!(r.is_err());
assert!(!slot.holds_write_lock());
assert!(slot.is_free_of_writer());
}
#[test]
fn with_exclusive_nests_and_keeps_a_lock_taken_at_open() {
let d = Scratch::new();
let slot = LockSlot::exclusive(&d.0).unwrap();
slot.with_exclusive::<(), io::Error>(|| {
slot.with_exclusive::<(), io::Error>(|| Ok(()))?;
assert!(slot.holds_write_lock(), "an inner section must not release the outer lock");
Ok(())
})
.unwrap();
assert!(slot.holds_write_lock(), "a slot opened exclusive keeps it");
}
#[test]
fn acquire_exclusive_promotes_a_shared_slot_and_is_idempotent() {
let d = Scratch::new();
let slot = LockSlot::shared(&d.0).unwrap();
slot.acquire_exclusive().unwrap();
slot.acquire_exclusive().unwrap();
assert!(slot.holds_write_lock());
}
/// The case a naive upgrade deadlocks on: two readers, each holding a shared
/// lock, both ask for the exclusive one at the same time.
#[test]
fn two_readers_promoting_at_once_both_finish() {
let d = Scratch::new();
let barrier = Arc::new(Barrier::new(2));
let (tx, rx) = mpsc::channel();
let handles: Vec<_> = (0..2)
.map(|_| {
let (dir, b, tx) = (d.0.clone(), Arc::clone(&barrier), tx.clone());
thread::spawn(move || {
let slot = LockSlot::shared(&dir).unwrap();
b.wait(); // both hold a shared lock
slot.with_exclusive::<(), io::Error>(|| {
assert!(slot.holds_write_lock());
thread::sleep(Duration::from_millis(50));
Ok(())
})
.unwrap();
tx.send(()).unwrap();
})
})
.collect();
for _ in 0..2 {
rx.recv_timeout(PATIENCE).expect("a promotion deadlocked");
}
for h in handles {
h.join().unwrap();
}
}
#[test]
fn a_lone_promotion_is_gapless_and_a_demotion_keeps_the_lock() {
let d = Scratch::new();
let shared = DirLock::acquire_shared(&d.0).unwrap();
let (excl, gapless) = shared.promote().unwrap();
assert_eq!(excl.mode(), LockMode::Exclusive);
assert!(gapless, "nobody else was promoting: the lock was never released");
let shared = excl.demote().unwrap();
assert_eq!(shared.mode(), LockMode::Shared);
// Shared again: other readers get in, a writer does not.
let _other = DirLock::acquire_shared(&d.0).unwrap();
assert!(DirLock::is_free_of_writer(&d.0).unwrap());
}
+3216
View File
File diff suppressed because it is too large. Load diff
+36
View File
@@ -0,0 +1,36 @@
[package]
name = "migrate-mphf"
version = "0.1.0"
edition = "2024"
publish = false
description = "One-shot converter: obikmer indexes whose layer MPHFs were built with ptr_hash 1.x -> ptr_hash 2.x"
# Standalone: not a member of src/'s workspace, own Cargo.lock. The previous
# generation of the MPHF stack (and its `proc-macro-error2` future-incompat
# warning) lives only here, never in obikmer itself.
[workspace]
[features]
default = []
numa = ["obisys/numa"]
[dependencies]
obikindex = { path = "../../src/obikindex" }
obicompactvec = { path = "../../src/obicompactvec" }
obiskio = { path = "../../src/obiskio" }
obisys = { path = "../../src/obisys", default-features = false }
epserde = "0.13"
rayon = "1"
anyhow = "1"
clap = { version = "4", features = ["derive"] }
tracing = "0.1.44"
tracing-subscriber = { version = "0.3", features = ["fmt", "env-filter"] }
# The generation the indexes to convert were built with, under other names
# so that both can live in this one binary.
ptr_hash_old = { package = "ptr_hash", version = "=1.1.0" }
epserde_old = { package = "epserde", version = "=0.8.0" }
cacheline_ef_old = { package = "cacheline-ef", version = "=1.1.0" }
[profile.release]
debug = 1
+98
View File
@@ -0,0 +1,98 @@
//! Crash-safe replacement of a layer's files by their converted versions.
//!
//! Everything new is written under `<layer>/.migrate_mphf/` first; nothing of
//! the layer is touched until that is complete and verified. Then a `COMMIT`
//! marker is written, and [`finish`] moves the staged files into place. Every
//! step of [`finish`] is idempotent ("if the staged item still exists, move
//! it"), so a layer interrupted after `COMMIT` is simply finished by the next
//! run. Files are only ever replaced by `rename`, never written in place:
//! some of them are hard-linked from elsewhere (`ls -l` link count 2), and an
//! in-place write would corrupt the other link.
use std::fs;
use std::io;
use std::path::Path;
use anyhow::{Context, Result};
pub const STAGING: &str = ".migrate_mphf";
pub const COMMIT: &str = "COMMIT";
/// Written last: the layer is converted, skip it on later runs.
pub const DONE: &str = ".mphf_migrated";
/// Where derived annexes go instead of being deleted.
pub const STALE_DIR: &str = "stale_annexes_pre_ptrhash2";
/// Replaced as directories.
const DIRS: [&str; 2] = ["presence", "counts"];
/// Replaced as files; `mphf.bin` last.
const FILES: [&str; 3] = ["evidence.bin", "fingerprint.bin", "mphf.bin"];
/// Slot-ordered or slot-derived annexes of a layer: wrong (silently) once the
/// slot numbering changes, so they are set aside, to be rebuilt.
const LAYER_ANNEXES: [&str; 7] = [
"siblings.psib",
"entropy.pent",
"entropy.noamb.pent",
"pairweight.ppwt",
"pairweight.noamb.ppwt",
"quartet.pbiv",
"quartet.noamb.pbiv",
];
/// Index-level annexes derived from the sibling annex.
pub const INDEX_ANNEXES: [&str; 2] = ["family_overlap.povl", "family_overlap.noamb.povl"];
/// Move the staged files into `dir`, set the stale annexes aside, mark the
/// layer done and drop the staging area.
pub fn finish(dir: &Path) -> Result<()> {
let staging = dir.join(STAGING);
for name in DIRS {
let staged = staging.join(name);
if staged.exists() {
let target = dir.join(name);
if target.exists() {
fs::remove_dir_all(&target).with_context(|| format!("removing {}", target.display()))?;
}
fs::rename(&staged, &target).with_context(|| format!("installing {}", target.display()))?;
}
}
for name in FILES {
let staged = staging.join(name);
if staged.exists() {
let target = dir.join(name);
fs::rename(&staged, &target).with_context(|| format!("installing {}", target.display()))?;
}
}
set_aside(dir, &LAYER_ANNEXES)?;
fs::write(dir.join(DONE), env!("CARGO_PKG_VERSION"))?;
fs::remove_dir_all(&staging)?;
Ok(())
}
/// Move each of `names` that exists in `dir` into `dir/STALE_DIR/`.
/// Returns how many were moved.
pub fn set_aside(dir: &Path, names: &[&str]) -> Result<usize> {
let aside = dir.join(STALE_DIR);
let mut moved = 0;
for name in names {
let from = dir.join(name);
if !from.exists() {
continue;
}
fs::create_dir_all(&aside)?;
fs::rename(&from, aside.join(name)).with_context(|| format!("setting aside {}", from.display()))?;
moved += 1;
}
Ok(moved)
}
/// Remove a staging area left by an interrupted, uncommitted run.
pub fn discard(dir: &Path) -> io::Result<()> {
match fs::remove_dir_all(dir.join(STAGING)) {
Err(e) if e.kind() != io::ErrorKind::NotFound => Err(e),
_ => Ok(()),
}
}
+100
View File
@@ -0,0 +1,100 @@
//! Permute the per-slot files of a layer into `staging/`.
//!
//! Presence and count matrices always come out in the compact row-major
//! (sparse) format, whatever they were: a dense source is first converted
//! sequentially (`build_from_dense`), then permuted from the sparse form,
//! whose rows are cheap to read at arbitrary slots.
use std::fs;
use std::path::Path;
use anyhow::{Result, ensure};
use obicompactvec::{
PersistentBitMatrix, PersistentIntMatrix, PersistentSparseBitMatrixBuilder,
PersistentSparseCompactIntMatrixBuilder,
};
use obikindex::layer::evidence::{Evidence, EvidenceWriter};
use obikindex::layer::fingerprint::{FingerprintVec, FingerprintVecWriter};
use crate::plan::Plan;
use crate::util::oki;
/// Every per-slot file of the layer in `dir` that exists, permuted into `staging`.
pub fn permute_all(dir: &Path, staging: &Path, plan: &Plan) -> Result<()> {
if dir.join("evidence.bin").exists() {
evidence(dir, staging, plan)?;
}
if dir.join("fingerprint.bin").exists() {
fingerprint(dir, staging, plan)?;
}
if dir.join("presence").is_dir() {
presence(dir, staging, plan)?;
}
if dir.join("counts").is_dir() {
counts(dir, staging, plan)?;
}
Ok(())
}
fn evidence(dir: &Path, staging: &Path, plan: &Plan) -> Result<()> {
let old = oki(Evidence::open(&dir.join("evidence.bin")))?;
ensure!(old.len() == plan.n, "evidence.bin has {} slots, layer has {}", old.len(), plan.n);
let mut out = EvidenceWriter::new(plan.n);
for (s, &o) in plan.new_to_old.iter().enumerate() {
let (chunk_id, rank) = old.decode(o as usize);
out.set(s, chunk_id, rank);
}
oki(out.write(&staging.join("evidence.bin")))
}
fn fingerprint(dir: &Path, staging: &Path, plan: &Plan) -> Result<()> {
let old = oki(FingerprintVec::open(&dir.join("fingerprint.bin")))?;
ensure!(old.n() == plan.n, "fingerprint.bin has {} slots, layer has {}", old.n(), plan.n);
let mut out = FingerprintVecWriter::new(plan.n, old.b());
for (s, &o) in plan.new_to_old.iter().enumerate() {
out.set(s, old.get(o as usize));
}
oki(out.write(&staging.join("fingerprint.bin")))
}
fn presence(dir: &Path, staging: &Path, plan: &Plan) -> Result<()> {
let out = staging.join("presence");
let src = match PersistentBitMatrix::open(dir)? {
// Mono-genome: no file, nothing is attached to slots.
PersistentBitMatrix::Implicit { .. } => return Ok(()),
PersistentBitMatrix::Sparse(s) => {
ensure!(s.n() == plan.n, "presence has {} rows, layer has {}", s.n(), plan.n);
PersistentSparseBitMatrixBuilder::build_permuted(&s, &plan.new_to_old, &out)?.close()?;
return Ok(());
}
dense => dense,
};
ensure!(src.n() == plan.n, "presence has {} rows, layer has {}", src.n(), plan.n);
let tmp = staging.join("presence_dense_to_sparse");
let sparse = PersistentSparseBitMatrixBuilder::build_from_dense(&src, &tmp)?.finish()?;
drop(src);
PersistentSparseBitMatrixBuilder::build_permuted(&sparse, &plan.new_to_old, &out)?.close()?;
drop(sparse);
fs::remove_dir_all(&tmp)?;
Ok(())
}
fn counts(dir: &Path, staging: &Path, plan: &Plan) -> Result<()> {
let out = staging.join("counts");
let src = match PersistentIntMatrix::open(dir)? {
PersistentIntMatrix::Sparse(s) => {
ensure!(s.n() == plan.n, "counts has {} rows, layer has {}", s.n(), plan.n);
PersistentSparseCompactIntMatrixBuilder::build_permuted(&s, &plan.new_to_old, &out)?.close()?;
return Ok(());
}
dense => dense,
};
ensure!(src.n() == plan.n, "counts has {} rows, layer has {}", src.n(), plan.n);
let tmp = staging.join("counts_dense_to_sparse");
let sparse = PersistentSparseCompactIntMatrixBuilder::build_from_dense(&src, &tmp)?.finish()?;
drop(src);
PersistentSparseCompactIntMatrixBuilder::build_permuted(&sparse, &plan.new_to_old, &out)?.close()?;
drop(sparse);
fs::remove_dir_all(&tmp)?;
Ok(())
}
Loaded 100 of 106 files, more files were not shown because too many files have changed in this diff. Show more