feat: implement persistent layered index and chunked binary format

Introduce the `obilayeredmap` specification and persistent MPHF-based index architecture for incremental multi-dataset indexing. Implement chunked binary serialization with a fixed `u8` k-mer count limit (256) and overlapping super-kmer segments. Add memory-mapped I/O and a companion `.idx` index file for allocation-free, O(1) unitig access. Update MkDocs navigation, enhance the k-mer comparison script, and add comprehensive tests for serialization, partitioning, and file I/O pipelines.

docmd/implementation/mphf.md

@@ -49,7 +49,7 @@ Build a new MPHF over the filtered kmer set only, with the exact key count avail
 
 **Phase 1** (provisional, discarded after spectrum computation): FMPHGO. Tolerates overestimated capacity, compact, no need to optimise for query speed on a temporary structure.
 
-**Phase 2** (persistent, queried repeatedly): open between FMPHGO and ptr_hash. Exact key count is available, so both operate optimally. ptr_hash's query speed advantage (2.1–3.3×) is meaningful for the persistent index but carries the risk of a very young crate. FMPHGO is the conservative default; ptr_hash is worth revisiting once it has broader production use.
+**Phase 2** (persistent, queried repeatedly): **ptr_hash**. Exact key count is available at phase 2, so ptr_hash operates optimally. Its query speed (≥2.1× over FMPHGO) and construction speed (≥3.1×) are meaningful for the persistent index; the space overhead at 2.4 bits/key is acceptable. The crate's youth (Feb 2025) was previously a concern; it is now accepted given the performance profile and the fact that each layer MPHF is independently rebuildable from its unitig file if needed.
 
 boomphf is effectively eliminated: its space overhead is the largest and its streaming-construction advantage does not apply here.
 
@@ -73,9 +73,64 @@ All three are in-memory structures. Their internal representation is flat bit ar
 
 No established Rust crate provides a natively on-disk MPHF. **SSHash** (Sparse and Skew Hash) is a complete kmer dictionary designed for disk access and is order-preserving (overlapping kmers receive consecutive indices → cache-friendly count access), but it is C++-only and covers more than just the MPHF layer.
 
+---
+
+## Multilayer index architecture
+
+### Motivation
+
+An index built from a single dataset A can be extended with a new dataset B without rebuilding. This supports incremental construction (adding species, samples, or sequencing runs) and enables set operations across heterogeneous sources.
+
+### Layer structure
+
+Each layer is a self-contained unit:
+
+```
+layer_i/
+  unitigs.bin     — packed 2-bit nucleotide sequences
+  mphf.bin        — ptr_hash index (phase-2, exact key count)
+  evidence.bin    — [(unitig_id, rank)] per MPHF slot  (see unitig_evidence.md)
+  counts.bin      — [u32] per MPHF slot
+```
+
+Layers are **disjoint**: a canonical kmer belongs to exactly one layer. Layer 0 is built from dataset A. Adding dataset B proceeds as follows:
+
+1. For each kmer in B: query layer 0 — if found, accumulate count into `counts_0[MPHF_0(kmer)]`.
+2. Collect all kmers of B not present in any existing layer → set `B \ A`.
+3. Build layer 1 from `B \ A` using the standard two-phase pipeline (spectrum, filter, ptr_hash).
+
+Adding a third dataset C repeats the process: probe layer 0, then layer 1, then build layer 2 from `C \ A \ B`.
+
+### Membership verification
+
+ptr_hash maps any input to a valid slot — it does not natively detect absent keys. Membership is verified using the evidence entry: decode the kmer from `(unitig_id, rank)` and compare to the query. A mismatch means the kmer is absent from this layer; probe the next layer.
+
+This makes the evidence layer load-bearing for correctness, not only for locality.
+
+### Query algorithm
+
+```
+fn query(kmer) → Option<count>:
+    for layer in layers:
+        slot = layer.mphf.query(kmer)
+        if layer.evidence.decode(slot) == kmer:
+            return Some(layer.counts[slot])
+    return None
+```
+
+Expected probe depth: 1 for kmers present in layer 0, increasing for rare kmers added in later layers. In practice, the dominant dataset (largest A) should be layer 0 to minimise average probe depth.
+
+### Layer count and probe cost
+
+Each probe is a ptr_hash lookup (~10 ns) plus one evidence decode (two array accesses). For L layers the worst case is L probes + 1 None. In practice L is small (2–5 for typical multi-species databases). No global data structure is needed to route queries; the layer chain is traversed in order.
+
+### Merging layers
+
+Two layer chains can be merged by re-indexing their union through the standard pipeline. This is expensive (full rebuild) but produces an optimal single-layer index. Merge is a maintenance operation, not a query-path requirement.
+
 ## Open questions
 
 - Confirm actual partition sizes and overestimation factor on representative metagenomic datasets.
-- Revisit ptr_hash for phase 2 once the crate has broader production track record.
-- Assess rkyv integration cost for FMPHGO if true zero-copy mmap becomes necessary for the persistent index.
+- **rkyv integration**: all flat arrays in a layer (evidence, counts, presence/absence matrix) map trivially to `rkyv::Archive` — fixed-size element types, no heap indirection. The presence/absence matrix is the strongest case: at 10 M kmers × 1 000 samples ≈ 1.25 GB per partition, zero-copy mmap via rkyv avoids loading the entire matrix at open time, letting the OS page cache serve only accessed pages. ptr_hash itself is internally a flat bit array and is structurally compatible with rkyv, but requires either native crate support or a wrapper. Assess the wrapper cost and whether ptr_hash is willing to adopt rkyv upstream.
 - Keep SSHash in mind if the indexing architecture is reconsidered at a higher level.
+- Determine optimal layer ordering heuristic (by kmer count? by query frequency?) for multi-species databases.

docmd/implementation/obilayeredmap.md

@@ -0,0 +1,168 @@
+# obilayeredmap — layered kmer index crate
+
+## Purpose
+
+`obilayeredmap` implements a persistent, incrementally extensible kmer index. The index is organised in three levels: **collection → partition → layer**. Each layer covers a disjoint kmer set (kmers absent from all earlier layers), wrapping a `ptr_hash` MPHF with associated per-slot data. Adding a new dataset never rebuilds existing layers.
+
+---
+
+## Three-level hierarchy
+
+```
+index_root/                        ← LayeredMap (collection)
+  meta.json
+  part_00000/                      ← Partition
+    layer_0/                       ← Layer
+      mphf.bin
+      unitigs.bin
+      evidence.bin
+      counts.bin
+      presence.bin
+    layer_1/
+      ...
+  part_00001/
+    layer_0/
+    layer_1/
+    ...
+```
+
+**Collection** (`index_root/`): global metadata — kmer size k, number of partitions, layer count, sample registry.
+
+**Partition** (`part_XXXXX/`): one directory per hash bucket. All kmers whose canonical minimiser hashes to bucket X land in `part_XXXXX`. Partitions are independent and can be processed in parallel. The partition count and routing scheme (minimiser → bucket) are fixed at collection creation and recorded in `meta.json`.
+
+**Layer** (`layer_N/`): within a partition, a layer is the MPHF and its associated data for one dataset addition. Layer 0 is built from the first dataset A; layer 1 covers kmers in B not present in layer 0; and so on. Layers within a partition are disjoint: each kmer belongs to exactly one layer.
+
+---
+
+## Layer file layout
+
+```
+layer_N/
+  mphf.bin            — ptr_hash MPHF (exact key count, phase-2 construction)
+  unitigs.bin         — packed 2-bit nucleotide sequences (concatenated, variable-length)
+  unitig_offsets.bin  — u32 per unitig: nucleotide offset of unitig j in unitigs.bin
+  evidence.bin        — u32 per MPHF slot: (unitig_id: 25 | rank: 7)
+  counts.bin          — u32 per MPHF slot (total kmer occurrences)
+  presence.bin        — bit matrix: n_slots × n_samples  [optional]
+```
+
+Unitigs have variable lengths. Each record in `unitigs.bin` is self-delimiting: it begins with a varint `seql` (sequence length in nucleotides) followed by `(seql+3)/4` packed bytes — the streaming format defined in `obiskio`. Sequential scan is always possible using the in-record `seql`.
+
+For O(1) random access, `unitig_offsets.bin` is a **precomputed derived index**: a u32 array of byte offsets into `unitigs.bin`, with n_unitigs + 1 entries (sentinel = total byte size). Built once at construction by a single sequential scan; reconstructible from `unitigs.bin` if lost. Access: `unitigs.bin[offsets[j] .. offsets[j+1]]`.
+
+All files except `mphf.bin` are flat arrays of fixed-size elements, serialised with **rkyv** for zero-copy mmap access. `mphf.bin` uses ptr_hash's native serialisation; rkyv integration is deferred (see open questions).
+
+### Evidence encoding
+
+Evidence maps each MPHF slot to its kmer's location in the unitig file. It serves two roles: membership verification (ptr_hash maps any input to a valid slot; decoding evidence and comparing to the query detects absent keys) and kmer reconstruction.
+
+```
+slot s  →  unitig_id: u25  |  rank: u7
+```
+
+Packed into a `u32` (29 bits used, 3 spare). Decoding:
+
+```
+kmer = unitigs[unitig_id][rank .. rank + k]   // 2-bit packed slice
+```
+
+`rank` is the kmer's 0-based index within the unitig (kmer units, not nucleotides). For k=31, m=11, the structural maximum is k − m + 1 = 21 kmers per unitig; the empirical maximum observed is ~46 kmers. A `u7` (0–127) is sufficient.
+
+### Presence/absence matrix
+
+Column-major bit matrix: column j (sample j) is a contiguous `n_slots`-bit array. This layout makes per-sample operations (union, intersection, diff over a column) cache-friendly. For large matrices (e.g. 10 M slots × 1 000 samples ≈ 1.25 GB per partition), rkyv + mmap avoids loading the full matrix at open time.
+
+---
+
+## Query path
+
+A kmer query routes through all three levels:
+
+1. **Partition routing**: hash canonical minimiser of the query kmer → partition index → open `part_XXXXX/`.
+2. **Layer probing**: iterate layers in order within the partition; for each layer compute `slot = mphf.query(kmer)`, then verify `evidence.decode(slot) == kmer`. First match wins.
+3. **Data access**: read `counts[slot]` and/or `presence[slot]` from the matching layer.
+
+```
+fn query(kmer) → Option<Hit>:
+    part = partition_of(kmer)
+    for (i, layer) in part.layers.iter().enumerate():
+        slot = layer.mphf.query(kmer)
+        if layer.evidence.decode(slot) == kmer:
+            return Some(Hit { layer: i, slot })
+    return None
+```
+
+Expected probe depth: 1 for kmers in layer 0, increasing for later layers. In practice the dominant dataset should be layer 0.
+
+---
+
+## Add-layer algorithm
+
+When adding dataset B to an existing index:
+
+1. For each partition, iterate kmers of B routed to that partition.
+2. Probe existing layers; collect kmers absent from all layers → `B \ index`.
+3. Build a new layer from `B \ index` using the two-phase pipeline (FMPHGO provisional → ptr_hash definitive).
+4. Append the new layer directory under each `part_XXXXX/`.
+5. Update `meta.json` (layer count, sample registry).
+
+Each partition's new layer is built independently; the operation is fully parallel across partitions.
+
+---
+
+## Core API (sketch)
+
+```rust
+// Open an existing index
+let map = LayeredMap::open(path)?;
+
+// Query a canonical kmer across all partitions and layers
+match map.query(kmer) {
+    Some(hit) => {
+        let count   = hit.count();
+        let present = hit.presence_row();  // bit slice over samples
+    }
+    None => { /* absent */ }
+}
+
+// Non-destructive extension with a new dataset
+// unitigs produced by the two-phase pipeline, one per partition
+let layer_idx = map.add_layer(unitigs_dir, counts_dir, presence_path)?;
+```
+
+---
+
+## Dependencies
+
+| crate | role |
+|---|---|
+| `ptr_hash` | phase-2 MPHF per layer |
+| `ph` (FMPHGO) | phase-1 provisional MPHF during layer construction |
+| `rkyv` | zero-copy serialisation of flat arrays (evidence, counts, presence) |
+| `memmap2` | mmap of layer files |
+| `bitm` | bit-packed presence matrix |
+
+---
+
+## Serialisation strategy
+
+All flat arrays use `rkyv::Archive`:
+
+```rust
+#[derive(Archive, Serialize, Deserialize)]
+struct Evidence { slots: Vec<u32> }   // packed (unitig_id: 25 | rank: 7)
+
+#[derive(Archive, Serialize, Deserialize)]
+struct Counts { data: Vec<u32> }
+```
+
+At open time, each file is mmapped and cast to its archived type — no allocation, no copy. The MPHF is loaded via ptr_hash's own API; a rkyv wrapper is a future refinement.
+
+---
+
+## Open questions
+
+- **ptr_hash + rkyv**: ptr_hash's internals are flat bit arrays; a rkyv-compatible wrapper is structurally feasible. Assess upstream willingness or implement a thin newtype wrapper.
+- **Presence matrix layout**: column-major favours per-sample operations; row-major favours per-kmer queries. Decide based on dominant access pattern.
+- **Layer merge**: merging two `LayeredMap` instances into a single-layer index requires full rebuild. Define API and cost model; maintenance operation, not query-path.
+- **Canonical kmer orientation**: evidence stores canonical kmer; strand recovery requires one 64-bit revcomp comparison at query time.

docmd/implementation/unitig_evidence.md

@@ -69,7 +69,9 @@ Consequence for `u8` capacity:
 | nucleotides | 255 nuc | 225 kmers |
 | **kmers** | **255 kmers** | **285 nuc** |
 
-On *Betula nana* (k=31, 256 partitions), m_u ≈ 37.9 kmers/unitig on average; no unitig length distribution data measured yet. The `rank` field (kmer index within the unitig) fits in a `u8` as long as no unitig exceeds 255 kmers — guaranteed by the split strategy below.
+**Structural maximum from superkmer construction.** For k=31 and m=11, the maximum number of consecutive kmers sharing the same minimiser is k − m + 1 = **21 kmers** (the minimiser traverses from position k−m to 0 as the window slides). A unitig that is a single full superkmer therefore has exactly 21 kmers. This is confirmed by a bimodal distribution in empirical data: a sharp peak at 21 kmers appears in all partitions, including the anomalous partition 145. The observed maximum is ~46 kmers (unitigs spanning more than one superkmer), well within u8 range.
+
+On *Betula nana* (k=31, 256 partitions), m_u ≈ 37.9 kmers/unitig on average. The `rank` field (kmer index within the unitig) fits in a `u8` as long as no unitig exceeds 255 kmers — guaranteed by the split strategy below and amply satisfied by empirical maximums (~46 kmers observed).
 
 ### Split strategy for long unitigs
 

mkdocs.yml

@@ -44,6 +44,7 @@ nav:
       - On-disk storage: implementation/storage.md
       - MPHF selection: implementation/mphf.md
       - Unitig evidence encoding: implementation/unitig_evidence.md
+      - obilayeredmap crate: implementation/obilayeredmap.md
   - Architecture:
       - Sequences: architecture/sequences/invariant.md
 

scripts/compare_kmers.py

@@ -75,7 +75,12 @@ def main():
     parser.add_argument("file_a", help="First FASTA file (reference)")
     parser.add_argument("file_b", help="Second FASTA file (to compare)")
     parser.add_argument(
-        "-k", "--kmer-size", type=int, default=31, metavar="K", help="k-mer size (default: 31)"
+        "-k",
+        "--kmer-size",
+        type=int,
+        default=31,
+        metavar="K",
+        help="k-mer size (default: 31)",
     )
     args = parser.parse_args()
 
@@ -104,6 +109,10 @@ def main():
 
     if only_a or only_b:
         print("\nSets differ.", file=sys.stderr)
+        if len(only_a) > 0 and len(only_b) <= 10:
+            print(f"\nOnly in A: {only_a}")
+        if len(only_b) > 0 and len(only_b) <= 10:
+            print(f"\nOnly in B: {only_b}")
         sys.exit(1)
     else:
         print("\nSets are identical.")

src/Cargo.lock

@@ -1615,7 +1615,10 @@ version = "0.1.0"
 dependencies = [
  "memmap2",
  "niffler 3.0.0",
+ "obikrope",
  "obikseq",
+ "obiread",
+ "obiskbuilder",
  "obiskio",
  "ph",
  "rayon",
@@ -1623,6 +1626,7 @@ dependencies = [
  "serde",
  "serde_json",
  "sysinfo 0.33.1",
+ "tempfile",
  "tracing",
 ]
 
@@ -1679,6 +1683,7 @@ name = "obiskio"
 version = "0.1.0"
 dependencies = [
  "lru",
+ "memmap2",
  "niffler 3.0.0",
  "obikseq",
  "rustix",

src/obikpartitionner/Cargo.toml

@@ -3,6 +3,13 @@ name = "obikpartitionner"
 version = "0.1.0"
 edition = "2024"
 
+[dev-dependencies]
+tempfile = "3"
+obikseq   = { path = "../obikseq",   features = ["test-utils"] }
+obiskbuilder = { path = "../obiskbuilder" }
+obiread   = { path = "../obiread" }
+obikrope  = { path = "../obikrope" }
+
 [dependencies]
 niffler = "3.0.0"
 remove_dir_all = "0.8"

src/obikpartitionner/src/partition.rs

@@ -616,3 +616,119 @@ impl Drop for KmerPartition {
         let _ = self.close();
     }
 }
+
+// ── integration tests ─────────────────────────────────────────────────────────
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use std::collections::HashMap;
+
+    use obikrope::Rope;
+    use obikseq::SuperKmer;
+    use obiskbuilder::build_superkmers;
+
+    const K: usize = 11;
+    const M: usize = 5;
+
+    fn setup() {
+        obikseq::params::set_k(K);
+        obikseq::params::set_m(M);
+    }
+
+    /// Direct canonical k-mer counts from ASCII sequences — ground truth.
+    fn direct_counts(seqs: &[&[u8]]) -> (u64, u64) {
+        let mut counts: HashMap<Vec<u8>, u64> = HashMap::new();
+        for seq in seqs {
+            for i in 0..seq.len().saturating_sub(K - 1) {
+                let km = SuperKmer::from_ascii(&seq[i..i + K]).to_ascii();
+                *counts.entry(km).or_insert(0) += 1;
+            }
+        }
+        let f0 = counts.len() as u64;
+        let f1: u64 = counts.values().sum();
+        (f0, f1)
+    }
+
+    /// Run the full pipeline on a list of sequences and return (f0, f1) from
+    /// the `kmer_spectrum_raw.json` produced by `count_partition`.
+    fn pipeline_counts(seqs: &[&[u8]]) -> (u64, u64) {
+        setup();
+
+        let mut rope_data: Vec<u8> = Vec::new();
+        for seq in seqs {
+            rope_data.extend_from_slice(seq);
+            rope_data.push(0x00);
+        }
+        let mut rope = Rope::new(None);
+        rope.push(rope_data);
+
+        let superkmers: Vec<_> = build_superkmers(rope, K, 1, 0.0);
+
+        let dir = tempfile::tempdir().unwrap();
+        let mut kp = KmerPartition::create(dir.path(), 0, K, M, true).unwrap();
+        kp.write_batch(superkmers).unwrap();
+        kp.close().unwrap();
+        kp.dereplicate().unwrap();
+
+        let part_dir = dir.path().join("part_00000");
+        let dedup_path = part_dir.join("dereplicated.skmer.zst");
+        if !dedup_path.exists() {
+            return (0, 0);
+        }
+        count_partition(&part_dir, &dedup_path, K).unwrap();
+
+        let spec: serde_json::Value = serde_json::from_reader(
+            fs::File::open(part_dir.join("kmer_spectrum_raw.json")).unwrap(),
+        ).unwrap();
+        let f0 = spec["f0"].as_u64().unwrap_or(0);
+        let f1 = spec["f1"].as_u64().unwrap_or(0);
+        (f0, f1)
+    }
+
+    #[test]
+    fn single_sequence_f0_f1_match() {
+        let seqs: &[&[u8]] = &[b"ACGTACGTACGTACGTACGT"];
+        let (ef0, ef1) = direct_counts(seqs);
+        let (gf0, gf1) = pipeline_counts(seqs);
+        assert_eq!(gf0, ef0, "f0 wrong: expected {ef0}, got {gf0}");
+        assert_eq!(gf1, ef1, "f1 wrong: expected {ef1}, got {gf1}");
+    }
+
+    #[test]
+    fn two_sequences_f0_f1_match() {
+        let seqs: &[&[u8]] = &[
+            b"ACGTACGTACGTACGTACGT",
+            b"TGCATGCATGCATGCATGCA",
+        ];
+        let (ef0, ef1) = direct_counts(seqs);
+        let (gf0, gf1) = pipeline_counts(seqs);
+        assert_eq!(gf0, ef0, "f0 wrong: expected {ef0}, got {gf0}");
+        assert_eq!(gf1, ef1, "f1 wrong: expected {ef1}, got {gf1}");
+    }
+
+    #[test]
+    fn repeated_sequence_f1_doubles() {
+        let seq = b"ACGTACGTACGTACGTACGT";
+        let seqs: &[&[u8]] = &[seq, seq];
+        let (ef0, ef1) = direct_counts(seqs);
+        let (gf0, gf1) = pipeline_counts(seqs);
+        assert_eq!(gf0, ef0, "f0 wrong: expected {ef0}, got {gf0}");
+        assert_eq!(gf1, ef1, "f1 wrong: expected {ef1}, got {gf1}");
+    }
+
+    #[test]
+    fn many_sequences_f0_f1_match() {
+        // 20 distinct sequences of length 40 — forces multiple super-kmers and
+        // multiple minimizer boundaries per sequence.
+        let bases = b"ACGT";
+        let seqs: Vec<Vec<u8>> = (0..20u32)
+            .map(|i| (0..40).map(|j| bases[((i * 7 + j * 3) % 4) as usize]).collect())
+            .collect();
+        let seq_refs: Vec<&[u8]> = seqs.iter().map(|v| v.as_slice()).collect();
+        let (ef0, ef1) = direct_counts(&seq_refs);
+        let (gf0, gf1) = pipeline_counts(&seq_refs);
+        assert_eq!(gf0, ef0, "f0 wrong: expected {ef0}, got {gf0}");
+        assert_eq!(gf1, ef1, "f1 wrong: expected {ef1}, got {gf1}");
+    }
+}

src/obikseq/src/lib.rs

@@ -21,6 +21,7 @@ pub mod unitig;
 
 pub use annotations::Annotation;
 pub use kmer::{CanonicalKmer, Kmer, Minimizer, hash_kmer};
+pub use packed_seq::MAX_KMERS_PER_CHUNK;
 pub use params::{k, m, set_k, set_m};
 pub use routable::RoutableSuperKmer;
 pub use sequence::Sequence;

src/obikseq/src/packed_seq.rs

@@ -22,6 +22,9 @@ 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;
+
 // ── PackedSeq ─────────────────────────────────────────────────────────────────
 
 /// 2-bit packed DNA sequence of arbitrary length ≥ 1.
@@ -229,22 +232,36 @@ impl PackedSeq {
         self.iter_kmers().map(|km| km.canonical())
     }
 
-    /// Serialise to a compact binary representation.
+    /// Extract nucleotides `[start, end)` as a new [`PackedSeq`]. Allocates.
+    pub fn sub(&self, start: usize, end: usize) -> Self {
+        debug_assert!(end > start && end <= self.seql());
+        let nucs: Vec<u8> = (start..end).map(|i| self.nucleotide(i)).collect();
+        Self::from_nucleotides(&nucs)
+    }
+
+    /// Serialise one chunk to binary.
     ///
-    /// Format: varint(seql) followed by raw packed bytes.
-    /// `tail` and `byte_len` are both derivable from `seql` and need not be stored.
+    /// Format: `[u8: n_kmers−1][packed bytes]`.
+    /// The caller must ensure `seql ≥ k` and `seql − k + 1 ≤ MAX_KMERS_PER_CHUNK`.
+    /// Use [`SuperKmer::write_to_binary`] for sequences that may exceed one chunk.
     pub fn write_to_binary<W: Write>(&self, w: &mut W) -> io::Result<()> {
-        write_varint(w, self.seql() as u64)?;
+        let k = crate::params::k();
+        let seql = self.seql();
+        debug_assert!(seql >= k, "sequence shorter than k");
+        debug_assert!(
+            seql - k + 1 <= MAX_KMERS_PER_CHUNK,
+            "chunk exceeds MAX_KMERS_PER_CHUNK; split before calling write_to_binary"
+        );
+        w.write_all(&[(seql - k) as u8])?;
         w.write_all(&self.seq)
     }
 
-    /// Deserialise from the compact binary format produced by [`write_to_binary`].
+    /// Deserialise one chunk from the binary format produced by [`write_to_binary`].
     /// Allocates exactly one `Box<[u8]>` for the packed bytes.
     pub fn read_from_binary<R: Read>(r: &mut R) -> io::Result<Self> {
-        let seql = read_varint(r)? as usize;
-        if seql == 0 {
-            return Err(io::Error::new(io::ErrorKind::InvalidData, "empty sequence"));
-        }
+        let mut buf = [0u8; 1];
+        r.read_exact(&mut buf)?;
+        let seql = buf[0] as usize + crate::params::k();
         let byte_len = (seql + 3) / 4;
         let tail = (seql % 4) as u8;
         let mut seq = vec![0u8; byte_len];

src/obikseq/src/superkmer.rs

@@ -12,7 +12,7 @@ use xxhash_rust::xxh3::xxh3_64;
 use crate::Annotation;
 use crate::Sequence;
 use crate::kmer::{CanonicalKmer, Kmer, KmerError};
-use crate::packed_seq::{PackedSeq, read_varint, write_varint};
+use crate::packed_seq::{MAX_KMERS_PER_CHUNK, PackedSeq, read_varint, write_varint};
 
 // ── SKAnnotation ──────────────────────────────────────────────────────────────
 
@@ -91,13 +91,37 @@ impl SuperKmer {
         Self { count: 1, inner }
     }
 
-    /// Serialise to compact binary. Format: varint(count) + varint((byte_len << 2) | tail) + bytes.
+    /// Serialise to compact binary: `[varint(count)][u8: n_kmers−1][packed bytes]` per chunk.
+    ///
+    /// Sequences with more than [`MAX_KMERS_PER_CHUNK`] kmers are transparently split into
+    /// overlapping chunks (k−1 nucleotide overlap, same count per chunk). Each chunk is an
+    /// independent, self-contained record — one [`read_from_binary`] call reads exactly one.
     pub fn write_to_binary<W: Write>(&self, w: &mut W) -> io::Result<()> {
-        write_varint(w, self.count as u64)?;
-        self.inner.write_to_binary(w)
+        let k = crate::params::k();
+        let seql = self.seql();
+        debug_assert!(seql >= k, "super-kmer shorter than k");
+        let n_kmers = seql - k + 1;
+        if n_kmers <= MAX_KMERS_PER_CHUNK {
+            write_varint(w, self.count as u64)?;
+            self.inner.write_to_binary(w)
+        } else {
+            let chunk_nucl = MAX_KMERS_PER_CHUNK + k - 1;
+            let stride = MAX_KMERS_PER_CHUNK;
+            let mut start = 0;
+            loop {
+                let end = (start + chunk_nucl).min(seql);
+                let mut chunk = self.inner.sub(start, end);
+                chunk.canonicalize();
+                write_varint(w, self.count as u64)?;
+                chunk.write_to_binary(w)?;
+                if end == seql { break; }
+                start += stride;
+            }
+            Ok(())
+        }
     }
 
-    /// Deserialise from the binary format produced by [`write_to_binary`].
+    /// Deserialise one chunk from the binary format produced by [`write_to_binary`].
     /// Allocates exactly one `Box<[u8]>` for the packed bytes.
     pub fn read_from_binary<R: Read>(r: &mut R) -> io::Result<Self> {
         let count = read_varint(r)? as u32;

src/obikseq/src/tests/superkmer.rs

@@ -67,27 +67,57 @@ fn seql_roundtrip() {
 
 // ── binary serialisation ──────────────────────────────────────────────────────
 
+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()
+}
+
 #[test]
 fn binary_roundtrip() {
-    for len in all_lengths() {
+    set_k(4);
+    let k = crate::params::k();
+    for len in binary_test_lengths(k) {
         let mut sk = SuperKmer::from_ascii(&make_seq(len));
         sk.set_count(42);
         let mut buf = Vec::new();
         sk.write_to_binary(&mut buf).unwrap();
         let sk2 = SuperKmer::read_from_binary(&mut buf.as_slice()).unwrap();
-        assert_eq!(
-            sk.to_ascii(),
-            sk2.to_ascii(),
-            "sequence mismatch for len={len}"
-        );
+        assert_eq!(sk.to_ascii(), sk2.to_ascii(), "sequence mismatch for len={len}");
         assert_eq!(sk2.count(), 42, "count mismatch for len={len}");
     }
 }
 
+#[test]
+fn binary_split_roundtrip() {
+    // A super-kmer > MAX_KMERS_PER_CHUNK kmers is split into multiple records on write.
+    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
+    let seql = MAX_KMERS_PER_CHUNK + k;
+    let mut sk = SuperKmer::from_ascii(&make_seq(seql));
+    sk.set_count(7);
+    let mut buf = Vec::new();
+    sk.write_to_binary(&mut buf).unwrap();
+    // Read all records back.
+    let mut slice = buf.as_slice();
+    let chunk0 = SuperKmer::read_from_binary(&mut slice).unwrap();
+    let chunk1 = SuperKmer::read_from_binary(&mut slice).unwrap();
+    assert!(slice.is_empty(), "unexpected trailing bytes");
+    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)
+}
+
 #[test]
 fn binary_packed_seq_roundtrip() {
     use crate::packed_seq::PackedSeq;
-    for len in all_lengths() {
+    set_k(4);
+    let k = crate::params::k();
+    for len in binary_test_lengths(k) {
         let ps = PackedSeq::from_ascii(&make_seq(len));
         let mut buf = Vec::new();
         ps.write_to_binary(&mut buf).unwrap();
@@ -98,7 +128,8 @@ fn binary_packed_seq_roundtrip() {
 
 #[test]
 fn binary_size_is_compact() {
-    // seql=4 (1 byte packed): varint(count=1, 1 byte) + varint((1<<2)|0=4, 1 byte) + 1 byte = 3 bytes
+    // ACGT with k=4: varint(count=1, 1 byte) + u8(n_kmers-1=0, 1 byte) + 1 packed byte = 3 bytes
+    set_k(4);
     let sk = SuperKmer::from_ascii(b"ACGT");
     let mut buf = Vec::new();
     sk.write_to_binary(&mut buf).unwrap();

src/obikseq/src/tests/unitig.rs

@@ -160,7 +160,12 @@ mod tests {
 
     #[test]
     fn binary_roundtrip_all_lengths() {
-        for len in test_lengths() {
+        // write_to_binary encodes a single chunk: seql must be in [k, MAX_KMERS_PER_CHUNK+k-1].
+        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();
+        for len in valid_lengths {
             let u = Unitig::from_ascii(&make_seq(len));
             let mut buf = Vec::new();
             u.write_to_binary(&mut buf).unwrap();

src/obiskbuilder/src/iter.rs

@@ -96,7 +96,7 @@ impl Iterator for SuperKmerIter<'_> {
             }
 
             // ── 1. Entropy check ─────────────────────────────────────────────
-            if self.stat.normalized_entropy().unwrap_or(1.0) <= self.theta {
+            if self.stat.normalized_entropy().unwrap_or(1.0) < self.theta {
                 let result = self.try_emit();
                 self.cursor.rewind(self.k - 1).ok();
                 self.reset();
@@ -168,6 +168,70 @@ mod tests {
     // k=11, m=5 — valeurs minimales du projet (k ∈ [11,31])
     const K: usize = 11;
 
+    /// Collect the set of canonical k-mers from a raw ASCII sequence (no NUL).
+    fn direct_canonical_kmers(seq: &[u8]) -> std::collections::HashSet<Vec<u8>> {
+        (0..seq.len().saturating_sub(K - 1))
+            .map(|i| obikseq::SuperKmer::from_ascii(&seq[i..i + K]).to_ascii())
+            .collect()
+    }
+
+    /// Collect the set of canonical k-mers emitted by SuperKmerIter over a rope.
+    fn iter_canonical_kmers(rope: &Rope) -> std::collections::HashSet<Vec<u8>> {
+        SuperKmerIter::new(rope, K, 1, 0.0)
+            .flat_map(|rsk| {
+                rsk.superkmer()
+                    .iter_canonical_kmers()
+                    .map(|km| km.to_ascii())
+                    .collect::<Vec<_>>()
+            })
+            .collect()
+    }
+
+    #[test]
+    fn coverage_single_segment() {
+        setup();
+        let seq = b"ACGTACGTACGTACGTACGT";
+        let rope = make_rope(&[seq.as_ref(), b"\x00"].concat());
+        let direct = direct_canonical_kmers(seq);
+        let from_iter = iter_canonical_kmers(&rope);
+        let missing: Vec<_> = direct.difference(&from_iter).collect();
+        assert!(
+            missing.is_empty(),
+            "k-mers perdus dans segment unique : {missing:?}"
+        );
+    }
+
+    #[test]
+    fn coverage_two_segments() {
+        setup();
+        let seg1 = b"ACGTACGTACGTACGTACGT";
+        let seg2 = b"TGCATGCATGCATGCATGCA";
+        let rope = make_rope(&[seg1.as_ref(), b"\x00", seg2.as_ref(), b"\x00"].concat());
+        let mut direct = direct_canonical_kmers(seg1);
+        direct.extend(direct_canonical_kmers(seg2));
+        let from_iter = iter_canonical_kmers(&rope);
+        let missing: Vec<_> = direct.difference(&from_iter).collect();
+        assert!(
+            missing.is_empty(),
+            "k-mers perdus dans deux segments : {missing:?}"
+        );
+    }
+
+    #[test]
+    fn coverage_minimizer_boundary() {
+        setup();
+        // sequence assez longue pour forcer plusieurs changements de minimiseur
+        let seq: Vec<u8> = (0..80).map(|i| b"ACGT"[i % 4]).collect();
+        let rope = make_rope(&[seq.as_slice(), b"\x00"].concat());
+        let direct = direct_canonical_kmers(&seq);
+        let from_iter = iter_canonical_kmers(&rope);
+        let missing: Vec<_> = direct.difference(&from_iter).collect();
+        assert!(
+            missing.is_empty(),
+            "k-mers perdus à la frontière de minimiseur : {missing:?}"
+        );
+    }
+
     #[test]
     fn single_segment_one_superkmer() {
         setup();

src/obiskio/Cargo.toml

@@ -10,6 +10,7 @@ lru = "0.12"
 serde = { version = "1", features = ["derive"] }
 serde_json = "1"
 
+memmap2 = "0.9"
 obikseq = { path = "../obikseq" }
 
 [dev-dependencies]

src/obiskio/src/codec.rs

@@ -17,63 +17,5 @@ pub(crate) fn read_superkmer<R: Read>(r: &mut R) -> io::Result<Option<SuperKmer>
 }
 
 #[cfg(test)]
-mod tests {
-    use super::*;
-    use std::io::Cursor;
-
-    fn make_sk(ascii: &[u8]) -> SuperKmer {
-        SuperKmer::from_ascii(ascii)
-    }
-
-    #[test]
-    fn roundtrip_single() {
-        let sk = make_sk(b"ACGTACGT");
-        let mut buf = Vec::new();
-        write_superkmer(&mut buf, &sk).unwrap();
-
-        let mut cur = Cursor::new(&buf);
-        let got = read_superkmer(&mut cur).unwrap().unwrap();
-        assert_eq!(sk.to_ascii(), got.to_ascii());
-        assert_eq!(sk.seql(), got.seql());
-    }
-
-    #[test]
-    fn roundtrip_all_lengths() {
-        let bases: Vec<u8> = (0..300).map(|i| b"ACGT"[i % 4]).collect();
-        let k = 11;
-        for len in (k..=k + 8).chain([255, 256, 257]) {
-            let sk = make_sk(&bases[..len]);
-            let mut buf = Vec::new();
-            write_superkmer(&mut buf, &sk).unwrap();
-
-            let mut cur = Cursor::new(&buf);
-            let got = read_superkmer(&mut cur).unwrap().unwrap();
-            assert_eq!(sk.to_ascii(), got.to_ascii(), "len={len}");
-            assert_eq!(sk.seql(), got.seql(), "len={len}");
-        }
-    }
-
-    #[test]
-    fn eof_returns_none() {
-        let buf: Vec<u8> = vec![];
-        let mut cur = Cursor::new(&buf);
-        assert!(read_superkmer(&mut cur).unwrap().is_none());
-    }
-
-    #[test]
-    fn multiple_records() {
-        let seqs: &[&[u8]] = &[b"AAAA", b"CCCC", b"GGGG", b"TTTT"];
-        let mut buf = Vec::new();
-        for s in seqs {
-            write_superkmer(&mut buf, &make_sk(s)).unwrap();
-        }
-
-        let mut cur = Cursor::new(&buf);
-        for s in seqs {
-            let got = read_superkmer(&mut cur).unwrap().unwrap();
-            let expected = make_sk(s);
-            assert_eq!(expected.to_ascii(), got.to_ascii());
-        }
-        assert!(read_superkmer(&mut cur).unwrap().is_none());
-    }
-}
+#[path = "tests/codec.rs"]
+mod tests;

src/obiskio/src/lib.rs

@@ -4,8 +4,10 @@ pub mod limits;
 pub mod meta;
 pub mod pool;
 pub mod reader;
+pub mod unitig_index;
 
 pub use error::{SKError, SKResult};
 pub use meta::SKFileMeta;
 pub use pool::{create_token, create_token_with, SKFilePool, SharedPool, SKFileWriter};
 pub use reader::{SKFileIter, SKFileReader};
+pub use unitig_index::{UnitigFileReader, UnitigFileWriter};

src/obiskio/src/pool.rs

@@ -428,229 +428,5 @@ impl Drop for SKFileWriter {
 // ── tests ──────────────────────────────────────────────────────────────────────
 
 #[cfg(test)]
-mod tests {
-    use super::*;
-    use crate::reader::SKFileReader;
-    use obikseq::{SuperKmer, set_k};
-    use tempfile::{NamedTempFile, TempDir};
-
-    const TEST_K: usize = 4;
-
-    fn make_sk(seed: usize) -> SuperKmer {
-        let bases: Vec<u8> = (0..8).map(|j| b"ACGT"[(seed + j) % 4]).collect();
-        SuperKmer::from_ascii(&bases)
-    }
-
-    fn pool(max_open: usize) -> SharedPool {
-        Arc::new(Mutex::new(SKFilePool::new(max_open)))
-    }
-
-    fn open_token(t: &mut SKFileWriter, sk: &SuperKmer) {
-        t.set_flush_threshold(1);
-        t.write(sk).unwrap(); // pending ≥ 1 → drain → fd opened
-    }
-
-    #[test]
-    fn creation_holds_no_fd() {
-        set_k(TEST_K);
-        let dir = TempDir::new().unwrap();
-        let p = pool(3);
-        for i in 0..10 {
-            create_token(&p, dir.path().join(format!("p{i}.zst"))).unwrap();
-        }
-        assert_eq!(p.lock().unwrap().open_count(), 0);
-    }
-
-    #[test]
-    fn pool_limits_open_fds() {
-        set_k(TEST_K);
-        let dir = TempDir::new().unwrap();
-        let p = pool(3);
-        let sk = make_sk(0);
-
-        let mut tokens: Vec<SKFileWriter> = (0..6)
-            .map(|i| create_token(&p, dir.path().join(format!("p{i}.zst"))).unwrap())
-            .collect();
-
-        for t in tokens.iter_mut() {
-            open_token(t, &sk);
-        }
-
-        assert!(
-            p.lock().unwrap().open_count() <= 3,
-            "open={}",
-            p.lock().unwrap().open_count()
-        );
-    }
-
-    #[test]
-    fn evicted_token_stays_logically_open() {
-        set_k(TEST_K);
-        let dir = TempDir::new().unwrap();
-        let p = pool(1);
-        let sk = make_sk(0);
-
-        let mut t0 = create_token(&p, dir.path().join("a.zst")).unwrap();
-        let mut t1 = create_token(&p, dir.path().join("b.zst")).unwrap();
-
-        open_token(&mut t0, &sk); // t0 fd open, pool full
-        open_token(&mut t1, &sk); // evicts t0, t1 fd open
-
-        assert!(t0.is_open(), "t0 must remain logically open after eviction");
-        assert_eq!(p.lock().unwrap().open_count(), 1);
-    }
-
-    #[test]
-    fn evicted_data_readable_after_close_all() {
-        set_k(TEST_K);
-        let dir = TempDir::new().unwrap();
-        let p = pool(1);
-        let sk = make_sk(0);
-
-        let mut t0 = create_token(&p, dir.path().join("a.zst")).unwrap();
-        let mut t1 = create_token(&p, dir.path().join("b.zst")).unwrap();
-
-        t0.set_flush_threshold(1);
-        t0.write(&sk).unwrap(); // t0 fd open, pool full
-        t1.set_flush_threshold(1);
-        t1.write(&sk).unwrap(); // evicts t0, t1 fd open
-
-        // t0 still has the record in pending (eviction just closed fd, pending stays in token)
-        // Actually: t0's pending was drained before drain() returned (drain clears pending).
-        // So t0 wrote its record, then was evicted (fd closed).
-
-        drop(t0);
-        drop(t1);
-        p.lock().unwrap().close_all().unwrap();
-
-        for name in &["a.zst", "b.zst"] {
-            let mut r = SKFileReader::open(dir.path().join(name)).unwrap();
-            let got = r.read_batch(10).unwrap();
-            assert_eq!(got.len(), 1, "{name}: expected 1 record");
-        }
-    }
-
-    #[test]
-    fn touch_moves_to_mru_so_lru_is_evicted() {
-        set_k(TEST_K);
-        let dir = TempDir::new().unwrap();
-        let p = pool(2);
-        let sk = make_sk(0);
-
-        let mut t0 = create_token(&p, dir.path().join("a.zst")).unwrap();
-        let mut t1 = create_token(&p, dir.path().join("b.zst")).unwrap();
-        let mut t2 = create_token(&p, dir.path().join("c.zst")).unwrap();
-
-        open_token(&mut t0, &sk); // t0 open
-        open_token(&mut t1, &sk); // t1 open, t0 LRU
-
-        // Write to t0 again → t0 becomes MRU, t1 becomes LRU
-        t0.set_flush_threshold(1);
-        t0.write(&sk).unwrap();
-
-        // Writing to t2 fills pool (cap=2) → evicts LRU = t1
-        open_token(&mut t2, &sk);
-
-        let open_count = p.lock().unwrap().open_count();
-        assert!(open_count <= 2, "open_count={open_count}");
-    }
-
-    #[test]
-    fn close_all_produces_readable_files() {
-        set_k(TEST_K);
-        let dir = TempDir::new().unwrap();
-        let p = pool(8);
-        let paths: Vec<_> = (0..4)
-            .map(|i| dir.path().join(format!("{i}.zst")))
-            .collect();
-
-        let mut tokens: Vec<SKFileWriter> = paths
-            .iter()
-            .map(|path| create_token(&p, path.clone()).unwrap())
-            .collect();
-
-        for (i, t) in tokens.iter_mut().enumerate() {
-            t.write(&make_sk(i)).unwrap();
-        }
-        // close tokens first so pending bytes are flushed and Zstd frames finalized
-        for t in tokens.iter_mut() {
-            t.close().unwrap();
-        }
-        p.lock().unwrap().close_all().unwrap();
-
-        for path in &paths {
-            let mut r = SKFileReader::open(path).unwrap();
-            let got = r.read_batch(10).unwrap();
-            assert_eq!(got.len(), 1);
-        }
-    }
-
-    #[test]
-    fn write_batch_roundtrip() {
-        set_k(TEST_K);
-        let dir = TempDir::new().unwrap();
-        let p = pool(4);
-        let sks: Vec<_> = (0..50).map(make_sk).collect();
-        let path = dir.path().join("batch.zst");
-
-        let mut t = create_token(&p, path.clone()).unwrap();
-        t.write_batch(&sks).unwrap();
-        t.close().unwrap();
-
-        let mut r = SKFileReader::open(&path).unwrap();
-        let got = r.read_batch(100).unwrap();
-        assert_eq!(got.len(), 50);
-        for (a, b) in sks.iter().zip(got.iter()) {
-            assert_eq!(a.to_ascii(), b.to_ascii());
-        }
-    }
-
-    #[test]
-    fn from_system_limits_bounded() {
-        set_k(TEST_K);
-        let pool = SKFilePool::from_system_limits();
-        assert!(pool.max_open() >= 16);
-        assert!(pool.max_open() <= MAX_POOL_SIZE);
-    }
-
-    #[test]
-    fn standalone_roundtrip_zstd() {
-        set_k(TEST_K);
-        let tmp = NamedTempFile::new().unwrap();
-        let sks: Vec<_> = (0..100).map(make_sk).collect();
-        {
-            let mut w = SKFileWriter::create(tmp.path()).unwrap();
-            w.write_batch(&sks).unwrap();
-            w.close().unwrap();
-        }
-        let mut r = SKFileReader::open(tmp.path()).unwrap();
-        let got = r.read_batch(200).unwrap();
-        assert_eq!(got.len(), 100);
-        for (a, b) in sks.iter().zip(got.iter()) {
-            assert_eq!(a.to_ascii(), b.to_ascii());
-        }
-    }
-
-    #[test]
-    fn standalone_close_prevents_write() {
-        set_k(TEST_K);
-        let tmp = NamedTempFile::new().unwrap();
-        let mut w = SKFileWriter::create(tmp.path()).unwrap();
-        w.close().unwrap();
-        assert!(!w.is_open());
-        assert!(w.write(&make_sk(0)).is_err());
-    }
-
-    #[test]
-    fn standalone_is_physically_open() {
-        set_k(TEST_K);
-        let tmp = NamedTempFile::new().unwrap();
-        let mut w = SKFileWriter::create(tmp.path()).unwrap();
-        assert!(!w.is_physically_open()); // fd deferred until first drain
-        w.set_flush_threshold(1);
-        w.write(&make_sk(0)).unwrap(); // triggers drain → fd opened
-        assert!(w.is_physically_open());
-        w.close().unwrap();
-        assert!(!w.is_physically_open());
-    }
-}
+#[path = "tests/pool.rs"]
+mod tests;

src/obiskio/src/reader.rs

@@ -143,70 +143,5 @@ impl Iterator for SKFileIter<'_> {
 }
 
 #[cfg(test)]
-mod tests {
-    use super::*;
-    use crate::pool::SKFileWriter;
-    use tempfile::NamedTempFile;
-
-    const TEST_K: usize = 4; // test sequences are 8 bases; k=4 gives n_kmers=5
-
-    fn setup() {
-        obikseq::params::set_k(TEST_K);
-    }
-
-    fn make_sks(n: usize) -> Vec<SuperKmer> {
-        (0..n)
-            .map(|i| {
-                let bases: Vec<u8> = (0..8).map(|j| b"ACGT"[(i + j) % 4]).collect();
-                SuperKmer::from_ascii(&bases)
-            })
-            .collect()
-    }
-
-    #[test]
-    fn iter_all() {
-        setup();
-        let tmp = NamedTempFile::new().unwrap();
-        let sks = make_sks(50);
-
-        {
-            let mut w = SKFileWriter::create(tmp.path()).unwrap();
-            w.write_batch(&sks).unwrap();
-        }
-
-        let mut r = SKFileReader::open(tmp.path()).unwrap();
-        let got: Vec<_> = r.iter().collect();
-        assert_eq!(got.len(), 50);
-        for (a, b) in sks.iter().zip(got.iter()) {
-            assert_eq!(a.to_ascii(), b.to_ascii());
-        }
-    }
-
-    #[test]
-    fn reopen_and_seek() {
-        setup();
-        let tmp = NamedTempFile::new().unwrap();
-        let sks = make_sks(20);
-
-        {
-            let mut w = SKFileWriter::create(tmp.path()).unwrap();
-            w.write_batch(&sks).unwrap();
-        }
-
-        let mut r = SKFileReader::open(tmp.path()).unwrap();
-        // Read 10, then simulate pool eviction + re-access
-        let first = r.read_batch(10).unwrap();
-        r.close();
-        r.reopen_and_seek().unwrap();
-        // Continue from position 10
-        let rest = r.read_batch(20).unwrap();
-        assert_eq!(first.len(), 10);
-        assert_eq!(rest.len(), 10);
-        for (a, b) in sks[..10].iter().zip(first.iter()) {
-            assert_eq!(a.to_ascii(), b.to_ascii());
-        }
-        for (a, b) in sks[10..].iter().zip(rest.iter()) {
-            assert_eq!(a.to_ascii(), b.to_ascii());
-        }
-    }
-}
+#[path = "tests/reader.rs"]
+mod tests;

src/obiskio/src/tests/codec.rs

@@ -0,0 +1,64 @@
+use super::*;
+use obikseq::set_k;
+use std::io::Cursor;
+
+fn make_sk(ascii: &[u8]) -> SuperKmer {
+    SuperKmer::from_ascii(ascii)
+}
+
+#[test]
+fn roundtrip_single() {
+    set_k(4);
+    let sk = make_sk(b"ACGTACGT");
+    let mut buf = Vec::new();
+    write_superkmer(&mut buf, &sk).unwrap();
+
+    let mut cur = Cursor::new(&buf);
+    let got = read_superkmer(&mut cur).unwrap().unwrap();
+    assert_eq!(sk.to_ascii(), got.to_ascii());
+    assert_eq!(sk.seql(), got.seql());
+}
+
+#[test]
+fn roundtrip_all_lengths() {
+    set_k(11);
+    let k: usize = 11;
+    let bases: Vec<u8> = (0..300).map(|i| b"ACGT"[i % 4]).collect();
+    // With k=11, seql=257 → n_kmers=247 ≤ 256: single chunk, no split.
+    for len in (k..=k + 8).chain([255, 256, 257]) {
+        let sk = make_sk(&bases[..len]);
+        let mut buf = Vec::new();
+        write_superkmer(&mut buf, &sk).unwrap();
+
+        let mut cur = Cursor::new(&buf);
+        let got = read_superkmer(&mut cur).unwrap().unwrap();
+        assert_eq!(sk.to_ascii(), got.to_ascii(), "len={len}");
+        assert_eq!(sk.seql(), got.seql(), "len={len}");
+    }
+}
+
+#[test]
+fn eof_returns_none() {
+    set_k(4);
+    let buf: Vec<u8> = vec![];
+    let mut cur = Cursor::new(&buf);
+    assert!(read_superkmer(&mut cur).unwrap().is_none());
+}
+
+#[test]
+fn multiple_records() {
+    set_k(4);
+    let seqs: &[&[u8]] = &[b"AAAA", b"CCCC", b"GGGG", b"TTTT"];
+    let mut buf = Vec::new();
+    for s in seqs {
+        write_superkmer(&mut buf, &make_sk(s)).unwrap();
+    }
+
+    let mut cur = Cursor::new(&buf);
+    for s in seqs {
+        let got = read_superkmer(&mut cur).unwrap().unwrap();
+        let expected = make_sk(s);
+        assert_eq!(expected.to_ascii(), got.to_ascii());
+    }
+    assert!(read_superkmer(&mut cur).unwrap().is_none());
+}

src/obiskio/src/tests/pool.rs

@@ -0,0 +1,217 @@
+use super::*;
+use crate::reader::SKFileReader;
+use obikseq::{SuperKmer, set_k};
+use tempfile::{NamedTempFile, TempDir};
+
+const TEST_K: usize = 4;
+
+fn make_sk(seed: usize) -> SuperKmer {
+    let bases: Vec<u8> = (0..8).map(|j| b"ACGT"[(seed + j) % 4]).collect();
+    SuperKmer::from_ascii(&bases)
+}
+
+fn pool(max_open: usize) -> SharedPool {
+    Arc::new(Mutex::new(SKFilePool::new(max_open)))
+}
+
+fn open_token(t: &mut SKFileWriter, sk: &SuperKmer) {
+    t.set_flush_threshold(1);
+    t.write(sk).unwrap();
+}
+
+#[test]
+fn creation_holds_no_fd() {
+    set_k(TEST_K);
+    let dir = TempDir::new().unwrap();
+    let p = pool(3);
+    for i in 0..10 {
+        create_token(&p, dir.path().join(format!("p{i}.zst"))).unwrap();
+    }
+    assert_eq!(p.lock().unwrap().open_count(), 0);
+}
+
+#[test]
+fn pool_limits_open_fds() {
+    set_k(TEST_K);
+    let dir = TempDir::new().unwrap();
+    let p = pool(3);
+    let sk = make_sk(0);
+
+    let mut tokens: Vec<SKFileWriter> = (0..6)
+        .map(|i| create_token(&p, dir.path().join(format!("p{i}.zst"))).unwrap())
+        .collect();
+
+    for t in tokens.iter_mut() {
+        open_token(t, &sk);
+    }
+
+    assert!(
+        p.lock().unwrap().open_count() <= 3,
+        "open={}",
+        p.lock().unwrap().open_count()
+    );
+}
+
+#[test]
+fn evicted_token_stays_logically_open() {
+    set_k(TEST_K);
+    let dir = TempDir::new().unwrap();
+    let p = pool(1);
+    let sk = make_sk(0);
+
+    let mut t0 = create_token(&p, dir.path().join("a.zst")).unwrap();
+    let mut t1 = create_token(&p, dir.path().join("b.zst")).unwrap();
+
+    open_token(&mut t0, &sk);
+    open_token(&mut t1, &sk);
+
+    assert!(t0.is_open(), "t0 must remain logically open after eviction");
+    assert_eq!(p.lock().unwrap().open_count(), 1);
+}
+
+#[test]
+fn evicted_data_readable_after_close_all() {
+    set_k(TEST_K);
+    let dir = TempDir::new().unwrap();
+    let p = pool(1);
+    let sk = make_sk(0);
+
+    let mut t0 = create_token(&p, dir.path().join("a.zst")).unwrap();
+    let mut t1 = create_token(&p, dir.path().join("b.zst")).unwrap();
+
+    t0.set_flush_threshold(1);
+    t0.write(&sk).unwrap();
+    t1.set_flush_threshold(1);
+    t1.write(&sk).unwrap();
+
+    drop(t0);
+    drop(t1);
+    p.lock().unwrap().close_all().unwrap();
+
+    for name in &["a.zst", "b.zst"] {
+        let mut r = SKFileReader::open(dir.path().join(name)).unwrap();
+        let got = r.read_batch(10).unwrap();
+        assert_eq!(got.len(), 1, "{name}: expected 1 record");
+    }
+}
+
+#[test]
+fn touch_moves_to_mru_so_lru_is_evicted() {
+    set_k(TEST_K);
+    let dir = TempDir::new().unwrap();
+    let p = pool(2);
+    let sk = make_sk(0);
+
+    let mut t0 = create_token(&p, dir.path().join("a.zst")).unwrap();
+    let mut t1 = create_token(&p, dir.path().join("b.zst")).unwrap();
+    let mut t2 = create_token(&p, dir.path().join("c.zst")).unwrap();
+
+    open_token(&mut t0, &sk);
+    open_token(&mut t1, &sk);
+
+    t0.set_flush_threshold(1);
+    t0.write(&sk).unwrap();
+
+    open_token(&mut t2, &sk);
+
+    let open_count = p.lock().unwrap().open_count();
+    assert!(open_count <= 2, "open_count={open_count}");
+}
+
+#[test]
+fn close_all_produces_readable_files() {
+    set_k(TEST_K);
+    let dir = TempDir::new().unwrap();
+    let p = pool(8);
+    let paths: Vec<_> = (0..4)
+        .map(|i| dir.path().join(format!("{i}.zst")))
+        .collect();
+
+    let mut tokens: Vec<SKFileWriter> = paths
+        .iter()
+        .map(|path| create_token(&p, path.clone()).unwrap())
+        .collect();
+
+    for (i, t) in tokens.iter_mut().enumerate() {
+        t.write(&make_sk(i)).unwrap();
+    }
+    for t in tokens.iter_mut() {
+        t.close().unwrap();
+    }
+    p.lock().unwrap().close_all().unwrap();
+
+    for path in &paths {
+        let mut r = SKFileReader::open(path).unwrap();
+        let got = r.read_batch(10).unwrap();
+        assert_eq!(got.len(), 1);
+    }
+}
+
+#[test]
+fn write_batch_roundtrip() {
+    set_k(TEST_K);
+    let dir = TempDir::new().unwrap();
+    let p = pool(4);
+    let sks: Vec<_> = (0..50).map(make_sk).collect();
+    let path = dir.path().join("batch.zst");
+
+    let mut t = create_token(&p, path.clone()).unwrap();
+    t.write_batch(&sks).unwrap();
+    t.close().unwrap();
+
+    let mut r = SKFileReader::open(&path).unwrap();
+    let got = r.read_batch(100).unwrap();
+    assert_eq!(got.len(), 50);
+    for (a, b) in sks.iter().zip(got.iter()) {
+        assert_eq!(a.to_ascii(), b.to_ascii());
+    }
+}
+
+#[test]
+fn from_system_limits_bounded() {
+    set_k(TEST_K);
+    let pool = SKFilePool::from_system_limits();
+    assert!(pool.max_open() >= 16);
+    assert!(pool.max_open() <= MAX_POOL_SIZE);
+}
+
+#[test]
+fn standalone_roundtrip_zstd() {
+    set_k(TEST_K);
+    let tmp = NamedTempFile::new().unwrap();
+    let sks: Vec<_> = (0..100).map(make_sk).collect();
+    {
+        let mut w = SKFileWriter::create(tmp.path()).unwrap();
+        w.write_batch(&sks).unwrap();
+        w.close().unwrap();
+    }
+    let mut r = SKFileReader::open(tmp.path()).unwrap();
+    let got = r.read_batch(200).unwrap();
+    assert_eq!(got.len(), 100);
+    for (a, b) in sks.iter().zip(got.iter()) {
+        assert_eq!(a.to_ascii(), b.to_ascii());
+    }
+}
+
+#[test]
+fn standalone_close_prevents_write() {
+    set_k(TEST_K);
+    let tmp = NamedTempFile::new().unwrap();
+    let mut w = SKFileWriter::create(tmp.path()).unwrap();
+    w.close().unwrap();
+    assert!(!w.is_open());
+    assert!(w.write(&make_sk(0)).is_err());
+}
+
+#[test]
+fn standalone_is_physically_open() {
+    set_k(TEST_K);
+    let tmp = NamedTempFile::new().unwrap();
+    let mut w = SKFileWriter::create(tmp.path()).unwrap();
+    assert!(!w.is_physically_open());
+    w.set_flush_threshold(1);
+    w.write(&make_sk(0)).unwrap();
+    assert!(w.is_physically_open());
+    w.close().unwrap();
+    assert!(!w.is_physically_open());
+}

src/obiskio/src/tests/reader.rs

@@ -0,0 +1,63 @@
+use super::*;
+use crate::pool::SKFileWriter;
+use tempfile::NamedTempFile;
+
+const TEST_K: usize = 4;
+
+fn setup() {
+    obikseq::params::set_k(TEST_K);
+}
+
+fn make_sks(n: usize) -> Vec<SuperKmer> {
+    (0..n)
+        .map(|i| {
+            let bases: Vec<u8> = (0..8).map(|j| b"ACGT"[(i + j) % 4]).collect();
+            SuperKmer::from_ascii(&bases)
+        })
+        .collect()
+}
+
+#[test]
+fn iter_all() {
+    setup();
+    let tmp = NamedTempFile::new().unwrap();
+    let sks = make_sks(50);
+
+    {
+        let mut w = SKFileWriter::create(tmp.path()).unwrap();
+        w.write_batch(&sks).unwrap();
+    }
+
+    let mut r = SKFileReader::open(tmp.path()).unwrap();
+    let got: Vec<_> = r.iter().collect();
+    assert_eq!(got.len(), 50);
+    for (a, b) in sks.iter().zip(got.iter()) {
+        assert_eq!(a.to_ascii(), b.to_ascii());
+    }
+}
+
+#[test]
+fn reopen_and_seek() {
+    setup();
+    let tmp = NamedTempFile::new().unwrap();
+    let sks = make_sks(20);
+
+    {
+        let mut w = SKFileWriter::create(tmp.path()).unwrap();
+        w.write_batch(&sks).unwrap();
+    }
+
+    let mut r = SKFileReader::open(tmp.path()).unwrap();
+    let first = r.read_batch(10).unwrap();
+    r.close();
+    r.reopen_and_seek().unwrap();
+    let rest = r.read_batch(20).unwrap();
+    assert_eq!(first.len(), 10);
+    assert_eq!(rest.len(), 10);
+    for (a, b) in sks[..10].iter().zip(first.iter()) {
+        assert_eq!(a.to_ascii(), b.to_ascii());
+    }
+    for (a, b) in sks[10..].iter().zip(rest.iter()) {
+        assert_eq!(a.to_ascii(), b.to_ascii());
+    }
+}

src/obiskio/src/tests/unitig_index.rs

@@ -0,0 +1,169 @@
+use super::*;
+use obikseq::{Kmer, Sequence as _, Unitig, set_k};
+use tempfile::tempdir;
+
+fn make_unitig(ascii: &[u8]) -> Unitig {
+    Unitig::from_ascii(ascii)
+}
+
+fn canonical_of(ascii: &[u8]) -> CanonicalKmer {
+    Kmer::from_ascii(ascii).unwrap().canonical()
+}
+
+fn write_read(seqs: &[&[u8]]) -> (tempfile::TempDir, UnitigFileReader) {
+    let dir = tempdir().unwrap();
+    let path = dir.path().join("unitigs.bin");
+    let mut w = UnitigFileWriter::create(&path).unwrap();
+    for s in seqs {
+        w.write(&make_unitig(s)).unwrap();
+    }
+    w.close().unwrap();
+    let r = UnitigFileReader::open(&path).unwrap();
+    (dir, r)
+}
+
+// ── I/O round-trip ────────────────────────────────────────────────────────────
+
+#[test]
+fn roundtrip_empty_index() {
+    set_k(4);
+    let dir = tempdir().unwrap();
+    let path = dir.path().join("unitigs.bin");
+    let w = UnitigFileWriter::create(&path).unwrap();
+    w.close().unwrap();
+    let r = UnitigFileReader::open(&path).unwrap();
+    assert_eq!(r.len(), 0);
+}
+
+#[test]
+fn roundtrip_unitigs() {
+    set_k(4);
+    let seqs: &[&[u8]] = &[b"ACGTACGT", b"TTTTCCCC", b"GGGAAA"];
+    let (_dir, r) = write_read(seqs);
+    assert_eq!(r.len(), seqs.len());
+    for (i, s) in seqs.iter().enumerate() {
+        assert_eq!(r.unitig(i), make_unitig(s), "unitig {i} mismatch");
+    }
+}
+
+// ── Bit extraction ────────────────────────────────────────────────────────────
+
+#[test]
+fn extract_kmer_raw_basic() {
+    // ACGT = 00 01 10 11 = 0x1B; k=4, j=0 → 0x1B << 56
+    let bytes = [0x1Bu8];
+    assert_eq!(extract_kmer_raw(&bytes, 0, 4), 0x1Bu64 << 56);
+}
+
+#[test]
+fn extract_kmer_raw_intra_byte_offset() {
+    // ACGT, j=1, k=3 → CGT = 01 10 11 = 0x1B (6 bits) → 0x1B << 58
+    let bytes = [0x1Bu8];
+    assert_eq!(extract_kmer_raw(&bytes, 1, 3), 0x1Bu64 << 58);
+}
+
+#[test]
+fn extract_kmer_raw_cross_byte() {
+    // Two bytes: ACGT | ACGT = [0x1B, 0x1B]
+    // j=3, k=4: nucleotides 3..7 = T A C G = 11 00 01 10 = 0b11000110 = 0xC6
+    let bytes = [0x1Bu8, 0x1Bu8];
+    assert_eq!(extract_kmer_raw(&bytes, 3, 4), 0xC6u64 << 56);
+}
+
+// ── revcomp / canonical ───────────────────────────────────────────────────────
+
+#[test]
+fn revcomp_palindrome() {
+    // ACGT is its own reverse complement
+    let raw = 0x1Bu64 << 56; // ACGT, k=4
+    assert_eq!(revcomp_raw(raw, 4), raw);
+}
+
+#[test]
+fn revcomp_asymmetric() {
+    // revcomp(TTTG) = CAAA
+    // TTTG = 11 11 11 10 = 0xFE → 0xFE << 56
+    // CAAA = 01 00 00 00 = 0x40 → 0x40 << 56
+    let tttg = 0xFEu64 << 56;
+    let caaa = 0x40u64 << 56;
+    assert_eq!(revcomp_raw(tttg, 4), caaa);
+    assert_eq!(revcomp_raw(caaa, 4), tttg);
+}
+
+#[test]
+fn canonical_raw_selects_minimum() {
+    let tttg = 0xFEu64 << 56;
+    let caaa = 0x40u64 << 56;
+    assert_eq!(canonical_raw(tttg, 4), caaa); // TTTG → canonical is CAAA
+    assert_eq!(canonical_raw(caaa, 4), caaa); // CAAA already canonical
+}
+
+// ── verify_canonical_kmer ─────────────────────────────────────────────────────
+
+#[test]
+fn verify_forward_canonical() {
+    // CAAA is canonical (< TTTG); stored forward in the unitig → direct match
+    set_k(4);
+    let (_dir, r) = write_read(&[b"CAAAACGT"]);
+    let query = canonical_of(b"CAAA");
+    assert!(r.verify_canonical_kmer(0, 0, query));
+}
+
+#[test]
+fn verify_reverse_complement_stored() {
+    // TTTG stored in the unitig; canonical form is CAAA
+    // verify must recognise the match despite the stored orientation being non-canonical
+    set_k(4);
+    let (_dir, r) = write_read(&[b"TTTGACGT"]);
+    let query = canonical_of(b"CAAA"); // == canonical_of(b"TTTG")
+    assert!(r.verify_canonical_kmer(0, 0, query));
+}
+
+#[test]
+fn verify_wrong_kmer_returns_false() {
+    set_k(4);
+    let (_dir, r) = write_read(&[b"TTTGACGT"]);
+    let wrong = canonical_of(b"AAAC");
+    assert!(!r.verify_canonical_kmer(0, 0, wrong));
+}
+
+#[test]
+fn verify_second_unitig_second_position() {
+    // Two unitigs; check kmer at j=1 of unitig 1 ("TTTGACGT")
+    // j=1 → nucleotides 1..5 = TTGA
+    set_k(4);
+    let (_dir, r) = write_read(&[b"ACGTACGT", b"TTTGACGT"]);
+    let query = canonical_of(b"TTGA");
+    assert!(r.verify_canonical_kmer(1, 1, query));
+}
+
+// ── Splitting ─────────────────────────────────────────────────────────────────
+
+#[test]
+fn short_unitig_not_split() {
+    // seql=259 → n_kmers=256 = MAX_KMERS_PER_CHUNK → no split
+    set_k(4);
+    let seq: Vec<u8> = (0..259_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);
+}
+
+#[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)
+    set_k(4);
+    let seq: Vec<u8> = (0..260_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(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])));
+}

src/obiskio/src/unitig_index.rs

@@ -0,0 +1,286 @@
+use std::fs::File;
+use std::io::{BufWriter, Write as _};
+use std::path::{Path, PathBuf};
+
+use memmap2::Mmap;
+use obikseq::{CanonicalKmer, Unitig};
+
+pub use obikseq::MAX_KMERS_PER_CHUNK;
+
+use crate::error::{SKError, SKResult};
+
+// ── Index file format ─────────────────────────────────────────────────────────
+//
+//   magic:           [u8; 4]  = b"UIDX"
+//   n_unitigs:       u32 LE
+//   seqls:           [u8; n_unitigs]       max kmer index per chunk (= n_kmers − 1)
+//   packed_offsets:  [u32; n_unitigs + 1]  byte offsets to packed bytes in the
+//                                          sequence file; last entry is sentinel
+//
+// Each sequence record in the binary file: [u8: n_kmers−1][packed bytes].
+// Offsets point to the first packed byte of each record, past the leading u8.
+// Unitigs with more than MAX_KMERS_PER_CHUNK kmers are transparently split by the
+// writer into overlapping chunks (k-1 nucleotide overlap) so no kmer is lost.
+
+const MAGIC: [u8; 4] = *b"UIDX";
+
+fn idx_path(path: &Path) -> PathBuf {
+    let mut s = path.as_os_str().to_owned();
+    s.push(".idx");
+    PathBuf::from(s)
+}
+
+// Extract a sub-sequence [start, end) nucleotides from a unitig.
+fn sub_unitig(unitig: &Unitig, start: usize, end: usize) -> Unitig {
+    unitig.sub(start, end)
+}
+
+// ── Writer ────────────────────────────────────────────────────────────────────
+
+/// Writes a sequence of [`Unitig`] to an uncompressed binary file and builds
+/// an offset index at close time.
+///
+/// Unitigs with more than [`MAX_KMERS_PER_CHUNK`] kmers are transparently split
+/// into overlapping chunks (k-1 nucleotide overlap) so no kmer is lost.
+///
+/// The companion index file (`path.idx`) is written on [`close`].
+/// The binary format per record is `[u8: n_kmers−1][packed 2-bit bytes]`.
+pub struct UnitigFileWriter {
+    path: PathBuf,
+    file: BufWriter<File>,
+    seqls: Vec<u8>,
+    packed_offsets: Vec<u32>,
+    next_offset: u32,
+    k: usize,
+}
+
+impl UnitigFileWriter {
+    pub fn create(path: &Path) -> SKResult<Self> {
+        let file = File::create(path).map_err(SKError::Io)?;
+        Ok(Self {
+            path: path.to_owned(),
+            file: BufWriter::new(file),
+            seqls: Vec::new(),
+            packed_offsets: Vec::new(),
+            next_offset: 0,
+            k: obikseq::params::k(),
+        })
+    }
+
+    /// Write a unitig, splitting it into chunks if it exceeds [`MAX_KMERS_PER_CHUNK`].
+    pub fn write(&mut self, unitig: &Unitig) -> SKResult<()> {
+        let seql = unitig.seql();
+        let k = self.k;
+
+        if seql < k {
+            return Ok(());
+        }
+
+        let n_kmers = seql - k + 1;
+        if n_kmers <= MAX_KMERS_PER_CHUNK {
+            return self.write_chunk(unitig);
+        }
+
+        // Split into overlapping chunks of MAX_KMERS_PER_CHUNK kmers.
+        // Overlap of k-1 nucleotides ensures no kmer is lost at boundaries.
+        let chunk_nucl = MAX_KMERS_PER_CHUNK + k - 1;
+        let stride = MAX_KMERS_PER_CHUNK;
+        let mut start = 0;
+        while start < seql {
+            let end = (start + chunk_nucl).min(seql);
+            self.write_chunk(&sub_unitig(unitig, start, end))?;
+            if end == seql {
+                break;
+            }
+            start += stride;
+        }
+        Ok(())
+    }
+
+    fn write_chunk(&mut self, unitig: &Unitig) -> SKResult<()> {
+        let seql = unitig.seql();
+        let byte_len = (seql + 3) / 4;
+
+        // Header is 1 byte (u8: n_kmers − 1 = seql − k); packed bytes follow.
+        self.packed_offsets.push(self.next_offset + 1);
+        self.seqls.push((seql - self.k) as u8);
+
+        unitig
+            .write_to_binary(&mut self.file)
+            .map_err(SKError::Io)?;
+
+        self.next_offset += 1 + byte_len as u32;
+        Ok(())
+    }
+
+    /// Flush the sequence file and write the companion `.idx`.
+    pub fn close(mut self) -> SKResult<()> {
+        self.file.flush().map_err(SKError::Io)?;
+        drop(self.file);
+
+        // Sentinel: byte offset past the last record's packed bytes.
+        let sentinel = match (self.packed_offsets.last(), self.seqls.last()) {
+            (Some(&last_off), Some(&last_seql)) => {
+                let seql = last_seql as u32 + self.k as u32;
+                last_off + (seql + 3) / 4
+            }
+            _ => 0,
+        };
+        self.packed_offsets.push(sentinel);
+
+        write_idx(&idx_path(&self.path), &self.seqls, &self.packed_offsets)
+    }
+
+    pub fn len(&self) -> usize {
+        self.seqls.len()
+    }
+
+    pub fn is_empty(&self) -> bool {
+        self.seqls.is_empty()
+    }
+}
+
+fn write_idx(path: &Path, seqls: &[u8], packed_offsets: &[u32]) -> SKResult<()> {
+    let mut w = BufWriter::new(File::create(path).map_err(SKError::Io)?);
+    w.write_all(&MAGIC).map_err(SKError::Io)?;
+    w.write_all(&(seqls.len() as u32).to_le_bytes()).map_err(SKError::Io)?;
+    w.write_all(seqls).map_err(SKError::Io)?;
+    for &off in packed_offsets {
+        w.write_all(&off.to_le_bytes()).map_err(SKError::Io)?;
+    }
+    w.flush().map_err(SKError::Io)
+}
+
+// ── Reader ────────────────────────────────────────────────────────────────────
+
+/// Read-only random-access view of a unitig file.
+///
+/// The sequence file is memory-mapped; the index is loaded into RAM on open.
+/// All per-kmer operations are O(1) and allocation-free.
+pub struct UnitigFileReader {
+    mmap: Mmap,
+    seqls: Vec<u8>,
+    packed_offsets: Vec<u32>,
+    k: usize,
+}
+
+impl UnitigFileReader {
+    pub fn open(path: &Path) -> SKResult<Self> {
+        let file = File::open(path).map_err(SKError::Io)?;
+        let mmap = unsafe { Mmap::map(&file).map_err(SKError::Io)? };
+        let (seqls, packed_offsets) = read_idx(&idx_path(path))?;
+        let k = obikseq::params::k();
+        Ok(Self { mmap, seqls, packed_offsets, k })
+    }
+
+    pub fn len(&self) -> usize {
+        self.seqls.len()
+    }
+
+    pub fn is_empty(&self) -> bool {
+        self.seqls.is_empty()
+    }
+
+    /// Return the nucleotide length of chunk `i`.
+    #[inline]
+    pub fn seql(&self, i: usize) -> usize {
+        self.seqls[i] as usize + self.k
+    }
+
+    /// Reconstruct chunk `i` as a [`Unitig`]. Allocates a copy of the packed bytes.
+    pub fn unitig(&self, i: usize) -> Unitig {
+        let seql = self.seqls[i] as usize + self.k;
+        let start = self.packed_offsets[i] as usize;
+        let byte_len = (seql + 3) / 4;
+        let tail = (seql % 4) as u8;
+        let bytes = self.mmap[start..start + byte_len].to_vec().into_boxed_slice();
+        Unitig::new(tail, bytes)
+    }
+
+    /// Extract the raw left-aligned u64 of the kmer at position `j` within chunk `i`.
+    #[inline]
+    pub fn raw_kmer(&self, i: usize, j: usize) -> u64 {
+        let start = self.packed_offsets[i] as usize;
+        extract_kmer_raw(&self.mmap[start..], j, self.k)
+    }
+
+    /// Return `true` iff the kmer at position `j` of chunk `i` equals `query`.
+    ///
+    /// O(1), zero allocation. The chunk may store either orientation of the kmer;
+    /// canonicalization is applied before comparison.
+    #[inline]
+    pub fn verify_canonical_kmer(&self, i: usize, j: usize, query: CanonicalKmer) -> bool {
+        canonical_raw(self.raw_kmer(i, j), self.k) == query.raw()
+    }
+}
+
+fn read_idx(path: &Path) -> SKResult<(Vec<u8>, Vec<u32>)> {
+    let data = std::fs::read(path).map_err(SKError::Io)?;
+    let mut pos = 0;
+
+    if &data[pos..pos + 4] != &MAGIC {
+        return Err(SKError::Io(std::io::Error::new(
+            std::io::ErrorKind::InvalidData,
+            "unitig index: bad magic",
+        )));
+    }
+    pos += 4;
+
+    let n = u32::from_le_bytes(data[pos..pos + 4].try_into().unwrap()) as usize;
+    pos += 4;
+
+    let seqls = data[pos..pos + n].to_vec();
+    pos += n;
+
+    let mut packed_offsets = Vec::with_capacity(n + 1);
+    for _ in 0..=n {
+        packed_offsets.push(u32::from_le_bytes(data[pos..pos + 4].try_into().unwrap()));
+        pos += 4;
+    }
+
+    Ok((seqls, packed_offsets))
+}
+
+// ── Kmer utilities ────────────────────────────────────────────────────────────
+
+/// Reverse complement of a left-aligned 2-bit kmer (same algorithm as [`KmerOf::revcomp`]).
+#[inline]
+fn revcomp_raw(raw: u64, k: usize) -> u64 {
+    let x = !raw;
+    let x = x.swap_bytes();
+    let x = ((x >> 4) & 0x0F0F0F0F0F0F0F0F) | ((x & 0x0F0F0F0F0F0F0F0F) << 4);
+    let x = ((x >> 2) & 0x3333333333333333) | ((x & 0x3333333333333333) << 2);
+    x << (64 - 2 * k)
+}
+
+/// Canonical form of a left-aligned 2-bit kmer: `min(kmer, revcomp(kmer))`.
+#[inline]
+fn canonical_raw(raw: u64, k: usize) -> u64 {
+    raw.min(revcomp_raw(raw, k))
+}
+
+// ── Bit extraction ────────────────────────────────────────────────────────────
+
+/// Extract the kmer at nucleotide position `j` from MSB-first 2-bit packed `bytes`.
+/// Returns a left-aligned u64 matching [`KmerOf`]'s internal representation.
+#[inline]
+fn extract_kmer_raw(bytes: &[u8], j: usize, k: usize) -> u64 {
+    let bit_start = j * 2;
+    let byte_start = bit_start / 8;
+    let bit_offset = bit_start % 8; // always 0, 2, 4, or 6
+    let bytes_needed = (bit_offset + 2 * k + 7) / 8; // ≤ 9 for k ≤ 32
+
+    let mut acc = 0u128;
+    for idx in 0..bytes_needed {
+        acc = (acc << 8) | bytes.get(byte_start + idx).copied().unwrap_or(0) as u128;
+    }
+
+    let shift = bytes_needed * 8 - bit_offset - 2 * k;
+    let mask = !0u64 >> (64 - 2 * k);
+    let raw = (acc >> shift) as u64 & mask;
+    raw << (64 - 2 * k)
+}
+
+#[cfg(test)]
+#[path = "tests/unitig_index.rs"]
+mod tests;