Extract dereplication logic into obikderep crate and centralize paths

Move partition dereplication logic into a dedicated `obikderep` crate that implements a two-phase hash-split/merge strategy with Rayon for parallel processing. Centralize superkmer file path construction in `obilayeredmap` and update dependent crates to use the new pipeline and shared path helpers. Adjust test suites to explicitly invoke the dereplication step.

DevDocMD/implementation/partition_layer_cache.md

@@ -8,8 +8,18 @@ state, added (panics on every read method). (2a) done — the
 onto it — **not started**, deliberately deferred. (3) done — the
 `obikindex ↔ obikpartitionner` dependency inverted: `PartitionRouter` now
 takes `&mut KmerIndex` and produces `Layer::Empty` shells directly, closing
-the gap `Layer::Empty` was built for in (1b) — see "(3) done" below.
+the gap `Layer::Empty` was built for in (1b) — see "(3) done" below. (4)
-Earlier mix-up, for
+done — dereplication split out into its own crate, `obikderep`, first step
+of an incremental "one algorithm at a time" split of `obikpartitionner`'s
+remaining bundle (`count_kmer`/`build_layers` not yet moved) — see "(4)
+done" below. **(5) — full design agreed, not yet implemented** —
+`KmerPartition` was found to be wired into nothing (`KmerIndex` never
+calls it; every path is still computed via free functions), and the fix
+turned out to be bigger than `KmerPartition` alone: `Layer`'s own
+constructors don't self-name either. Full redesign of both, agreed in
+detail, session ended (budget) before implementation — see "(5) design
+agreed" below; **read it before touching `KmerPartition`/`Layer`
+signatures**, the shape is fully specified. Earlier mix-up, for
 context: an earlier
 version of this doc used the name `KmerPartition` (singular) for what was
 actually the *collection* type (later renamed `KmerPartitions`, later
@@ -447,6 +457,289 @@ Full workspace suite green (`cargo check --workspace --all-targets` +
 `index_layer.rs` fix above — the smoke test is what actually caught the
 regression the suite missed.
 
+## (4) done (2026-08-20): `obikderep` — dereplication split out of `obikpartitionner`, one algorithm at a time
+
+Follow-on question after (3): the indexing pipeline has 4 stages (scatter,
+dereplicate, count_kmer, index-build — see the CLI's own `Reporter` output,
+one line per stage), but `obikpartitionner` — a name that says
+*partitioning* — owned three of them (routing, dereplication, counting).
+Challenged directly, same as (3)'s dependency-direction question: a crate
+should hold what its name says, not accumulate unrelated stages just
+because they happened to land there first. Two ways to fix it — one crate
+renamed to hold all remaining stages, or one crate per stage — decided in
+favour of the latter, explicitly **incremental**: build the *second* algo
+crate first (`obikderep`, dereplication only), only then look at what it
+and `PartitionRouter` actually have in common, and factor a shared
+`Algorithm` trait (future `obikalgorithm` crate) from that real overlap —
+not guessed at from a single example. `count_kmer` and `build_layers`
+(currently `KmerIndex` inherent methods — itself flagged as inconsistent
+with "`KmerIndex` is a data structure, not a compute structure") are left
+alone this round, on purpose — one stage moves at a time.
+
+**`obikderep`** (new crate): `Dereplicator<'a> { index: &'a KmerIndex, n_partitions, level }`
+— `new(index: &KmerIndex)` (shared borrow, not `&mut`: dereplication never
+writes index metadata), no setters yet (nothing to configure), `run(on_progress)`
+does the two-phase split+merge dereplication in parallel across partitions,
+ported unchanged from `PartitionRouter::dereplicate` (moved wholesale:
+`optimal_buckets`/`dereplicate_partition`/`load_bucket`/`flush_map`/
+`remove_skmer_file`, now private to this crate in `dereplicate.rs`).
+`obikpartitionner::PartitionRouter::dereplicate` is gone; `count_kmer`
+stays.
+
+**A real signature difference from `PartitionRouter::run`, not an
+inconsistency**: `Dereplicator::run` takes `Option<impl Fn(Progress) +
+Sync>`, not `FnMut`. `PartitionRouter::run`'s callback is invoked from one
+sequential loop (`FnMut` is fine); `Dereplicator::run`'s work is
+`rayon::par_iter`, so the callback can be invoked concurrently from
+multiple worker threads — same reason `obisys::TracedBar`'s own methods
+take `&self`, not `&mut self`. Progress position is tracked with an
+`AtomicU64`, incremented from inside the parallel closure so each
+completed partition reports immediately — collecting all results first and
+reporting after (the first draft of this) would have delivered every tick
+in one burst at the very end, defeating the point of a live progress bar.
+`total: Some(n_partitions)` (known up front, unlike scatter's bases count)
+— `cmd/index/mod.rs` renders a real `progress_bar`, not a spinner, driven
+by the callback exactly like scatter's spinner is.
+
+**A second, pre-existing instance of the exact bug (3) fixed, caught
+before it shipped**: `dereplicated.skmer.zst` was hand-built as a string
+literal independently in three places — `obikpartitionner`'s
+`dereplicate.rs`/`count.rs` *and* `obikindex`'s `index_layer.rs` (a literal
+that already predated this session, never caught until now). Splitting
+dereplication into its own crate turns this from "two places, still
+matching by luck" into "three independent crates that must agree on a
+filename with no shared dependency forcing them to" — no longer
+deferrable. Fixed by adding `obilayeredmap::{raw_superkmers_path,
+dereplicated_superkmers_path}` (free functions, `layer_dir: &Path ->
+PathBuf`, mirroring `layer_dir` itself) — the filename lives in one place,
+in the Layer-tier crate every consumer here already depends on
+(`obikpartitionner`, `obikderep`, `obikindex` all reach it without a new
+edge), and no external crate ever sees the literal `"skmer.zst"` again.
+This reverses (3)'s own earlier call to keep `SK_EXT` private to
+`obikpartitionner` — that call assumed a single owner; a second owner
+appearing (`obikderep`) removed the assumption it rested on, so the
+decision changed with it, not out of inconsistency.
+
+Every `count_kmer` call site that used to run after `router.dereplicate()`
+on the same `PartitionRouter` now runs after a separate
+`Dereplicator::new(&idx).run(...)` call, on a freshly-constructed
+`PartitionRouter` — `PartitionRouter` no longer offers a combined
+"dereplicate then count" path. Updated at all three call sites that had
+one: `cmd/index/mod.rs`, `obikphylo`'s test harness, `obikpartitionner`'s
+own tests.
+
+Full workspace suite green (`cargo check --workspace --all-targets` +
+`cargo test --workspace`, exit code 0), plus an end-to-end CLI smoke test
+against real FASTA data (scatter → dereplicate → count → index-build →
+query, same numbers as (3)'s smoke test: 870 kmers) — required this time
+too, per (3)'s own lesson: no test in the suite exercises `obikmer index`'s
+real file-driven path.
+
+Still not done: `count_kmer`/`build_layers` staying where they are, the
+`obikalgorithm` shared-trait extraction (deliberately deferred until a
+third data point exists), and everything already listed under (2b).
+
+## (5) design agreed, not yet implemented (2026-08-20): `KmerPartition` rewritten, `Layer` gains self-naming, a future cache crate over `KmerIndex`
+
+Session ended (out of budget) before any of this was coded. Everything
+below is a **fully specified plan**, agreed sentence by sentence with the
+user — not a sketch to re-derive, not a proposal to re-litigate. Implement
+it as written; if something here turns out to be wrong once coded, fix it
+and update this section, don't restart the design conversation.
+
+### How this was found
+
+Direct question from the user: "tu as bien créé une structure
+`KmerPartition` ?" — yes (2a), but investigating exposed that it is
+**wired into nothing**. `KmerIndex` has no `partition(i)` method at all;
+`partition_dir`/`index_dir`/`layer_dir` still call `obikpartition::
+partition_dir`/`index_dir` and `obilayeredmap::layer_dir` as bare free
+functions directly, never touching a `KmerPartition`/`Layer` object to get
+there. The end result on disk is identical (same paths), which is exactly
+why no test caught it — but the *responsibility* is in the wrong place:
+one function (on `KmerIndex`) knows the whole three-tier naming
+convention, instead of each tier asking the one below it for its own
+path. User's framing, verbatim, now saved as [[feedback_no_spaghetti_petits_pois]]:
+"spaghetti" (logic untraceable, split across too many unrelated crates)
+and "petits pois" (small bits of naming logic dispersed with no owning
+object) are **strictly forbidden** — this was a live example of both.
+
+Pushed further, twice:
+1. First correction: `Layer::create(&obilayeredmap::layer_dir(&dir, 0))` —
+   still a free-function call from *outside* `Layer` to compute where it
+   should live. "Le layer n'est pas con, c'est lui qui dit où est-ce qu'il
+   doit être sauvé" (the layer isn't stupid, it says itself where it
+   should be saved).
+2. Second correction, the general principle: **"une partition est juste
+   identifiée par un numéro, tout se calcule à partir du numéro, et un
+   layer est identifié à partir d'un numéro et tout se calcule à partir de
+   ce numéro."** Concretely: each object stores its own local identifying
+   number *plus* its immediate parent's path (captured once, at
+   construction) — never a path handed in again later by a caller, and
+   never a free function outside the object that can compute that path
+   independently. Explicitly rejected along the way: making users pass
+   "the partition's path that contains the layer" to open a layer — the
+   parent path is captured once, at the child's construction, not
+   re-supplied at every call.
+
+### The agreed shape
+
+**`Layer`** (`obilayeredmap`) — identified by `l` + its parent partition's
+directory, both captured at construction, never received again:
+
+```rust
+pub enum Layer {
+    Empty { partition_dir: PathBuf, l: usize },   // pure identification, no disk I/O
+    Count(TypedLayer<PersistentCompactIntMatrix>),
+    Presence(TypedLayer<PersistentBitMatrix>),
+}
+
+impl Layer {
+    pub fn at(partition_dir: &Path, l: usize) -> Self;                 // identify only
+    fn dir(&self) -> PathBuf;                                          // private — layer_dir() no longer a public free function, folded in here
+    pub fn create(self) -> io::Result<Self>;                           // creates the directory if needed; no path parameter anymore
+    pub fn open(self, mode: &IndexMode, with_counts: bool) -> OLMResult<Self>;  // no path parameter anymore
+    // mphf_path()/unitigs_path()/evidence_path()/fingerprint_path()/counts_dir()/presence_dir()
+    // unchanged in spirit, implemented via self.dir() instead of a stored `dir` field read directly
+}
+```
+
+Note this **replaces** `Layer::Empty { dir: PathBuf }` from (1b) — `dir`
+becomes a computed value (`partition_dir.join(format!("layer_{l}"))`), not
+a stored field. `obilayeredmap::layer_dir`/`raw_superkmers_path`/
+`dereplicated_superkmers_path` (currently public free functions,
+introduced in (3)/(4)) stop being called from outside `obilayeredmap`
+entirely once this lands — they were the right fix for their moment (a
+second crate, `obikderep`, needed to agree on a filename with no owner),
+but the *real* fix, now visible with a third data point, is that `Layer`
+itself should be the only thing anyone asks.
+
+**`KmerPartition`** (`obikpartition`) — same principle, one tier up:
+
+```rust
+pub struct KmerPartition {
+    index_root: PathBuf,   // the parent KmerIndex's root, captured once
+    i: usize,
+}
+
+impl KmerPartition {
+    pub fn new(index_root: PathBuf, i: usize) -> Self;                 // identify only, no disk I/O
+    pub fn create(index_root: PathBuf, i: usize) -> io::Result<Self>;  // creates this partition's directory + an empty layer 0 (a partition is never born without one — that knowledge lives here, not in whoever calls create)
+    pub fn partition_dir(&self) -> PathBuf;                            // part_{i:05}
+    pub fn index_dir(&self) -> PathBuf;                                // part_{i:05}/index
+    pub fn layer(&self, l: usize) -> Layer;                            // Layer::at(&self.index_dir(), l) — caller never touches a path
+    pub fn meta(&self) -> SKResult<PartitionMeta>;                     // n_layers + mode; must absorb the recovery-on-missing-file logic
+                                                                         // currently private in obikindex::common::load_meta (obikpartition
+                                                                         // can't depend on obikindex to reuse it — this logic moves down)
+    pub fn n_layers(&self) -> SKResult<usize>;                         // meta()?.n_layers
+    pub fn mode(&self) -> SKResult<IndexMode>;                         // meta()?.mode — "exact/approximatif"
+    pub fn is_filled(&self) -> bool;                                   // does this partition's directory exist at all
+    pub fn n_kmers(&self) -> io::Result<usize>;                        // LayerMeta::load(&self.layer(0).dir()).n — reads layer 0's count as
+                                                                         // a representative figure, same "read the first one" trick
+                                                                         // n_layers_per_partition() already uses at the KmerIndex level
+}
+```
+
+This **replaces** (2a)'s `KmerPartition { layers: Vec<Layer> }` entirely
+— no eagerly-opened `Vec<Layer>`, no `find()` (both belong to the future
+cache, see below, which is the thing that actually holds opened layers
+alive across many lookups). (2a)'s version is safe to delete outright: it
+was never wired into anything (confirmed above), so nothing depends on
+its current shape. New dependencies needed: `obikpartition` gains
+`obiskio` (for `SKResult`) and `obicompactvec` (for `LayerMeta`).
+
+Deliberately **not built this round**: cross-level consistency checks
+("verify everything below me is in the same state") — a real idea, raised
+by the user, but nothing concrete needs it yet; building it speculatively
+would be exactly the premature-abstraction pattern this project avoids.
+
+**`KmerIndex`** (`obikindex`) — becomes the sole entry point:
+
+```rust
+pub fn partition(&self, i: usize) -> KmerPartition;   // KmerPartition::new(self.root_path.clone(), i)
+```
+
+`partition_dir(i)`/`index_dir(i)`/`layer_dir(i, l)` **stay** as public
+methods (≈30 existing call sites across `obikindex`/`obikphylo` — see (3)'s
+option A, applied identically here) but become pure delegations:
+`self.partition(i).partition_dir()`, `self.partition(i).index_dir()`,
+`self.partition(i).layer(l).dir()` (needs `Layer::dir()` to be visible
+enough for this — likely `pub(crate)` in `obilayeredmap` plus a thin
+public wrapper, or a public accessor on `Layer` itself; not fully nailed
+down, decide while implementing). No caller outside `obikindex` changes.
+
+### Known blast radius (why this wasn't done in the same session)
+
+- **14 files** call `Layer::open`/`Layer::create` directly today
+  (`obikphylo/siblings/{cache,build,family_scan,tests}.rs`,
+  `obikpartitionner/partition/router.rs`, `obikpartition/src/lib.rs`,
+  `obikindex/{rebuild_layer,dump_layer,index,query_layer}.rs`,
+  `obilayeredmap/{mphf_layer,layer,map,content_layer}.rs`) — every one
+  loses its path parameter and gains a `(partition_dir, l)` or an
+  already-identified `Layer` to call `.create()`/`.open()` on instead.
+- **≈30 files** call `KmerIndex::partition_dir`/`index_dir`/`layer_dir` —
+  unaffected in their own code (same public signatures), but worth
+  re-checking once (5) lands that none of them were relying on the old
+  free-function-based implementation in a way the new delegation breaks.
+- `obikpartitionner::PartitionRouter::ensure_writer` and `obikderep`'s
+  `run` both currently call `obilayeredmap::{layer_dir, raw_superkmers_path,
+  dereplicated_superkmers_path}` directly (from (3)/(4)) — both need to
+  switch to going through `index.partition(i).layer(0)` instead.
+
+### Also agreed, separately: `PartitionRouter::new` never needed `&mut KmerIndex`
+
+Verified by reading the code: every call `PartitionRouter` makes on
+`index` is `&self` (`index.kmer_size()`, `index.index_dir(i)`). The `&mut`
+in its current signature (from (3)) was inherited from the original
+"the router writes to the partitions" reasoning, never actually required
+by any method call. This is *exactly* what caused every `drop(router)`
+workaround needed throughout (3)/(4) (`cmd/index/mod.rs` ×2, `obikphylo`'s
+test harness, `obikpartitionner`'s own tests) — `PartitionRouter` holds a
+`Drop` impl, which extends a `&mut` borrow to the end of its scope even
+past its last real use. **Fix alongside (5)**: change
+`PartitionRouter::new(index: &'a mut KmerIndex)` to `&'a KmerIndex`, and
+remove the now-unnecessary `drop(router)` calls at all four sites.
+
+### Also discussed: a future cache crate, not part of (5), not `obikalgorithm` either
+
+Separate idea, explicitly **not** part of this design and **not** started:
+a new crate whose only job is to cache open `KmerPartition`s (and their
+opened `Layer`s) across one run — replacing both `obikphylo::siblings::
+cache::PartitionCache` (today, sibling-specific, holds `Vec<Vec<Layer>>`)
+and `obikindex::query_layer::QueryLayer` (today, uncached, bypasses
+`Layer` entirely) — the two consumers (2b) already identified as each
+reinventing a fragment of the same thing.
+
+User's framing: this is **not** a third `obikalgorithm` data point — an
+algorithm has a `new → run → done` shape; a cache has a fundamentally
+different one (open, stay alive for a whole run, serve lookups, maybe
+evict) — "on crée un cache sur un index, ça consomme un index." Two
+distinct crate *roles* in this ecosystem (data crates: `obikpartition`/
+`obilayeredmap`; algorithm crates: `obikpartitionner`/`obikderep`/future
+`obikalgorithm` implementors; and now a cache/service crate), not one
+unified shape to force everything into.
+
+Depends on (5) being done first: the cache crate's whole job is holding
+`Vec<KmerPartition>`/opened `Layer`s alive, built via `KmerIndex::
+partition(i)` as its factory — nothing to build it on top of until (5)
+lands. Still open once (5) is done: eviction policy vs. holding everything
+open for the process lifetime (the never-measured mmap/VM-mapping-count
+question from earlier in this doc), and whether it lives in `obikpartition`
+itself or a new crate.
+
+### Order of remaining work, as currently understood
+
+1. **(5)** — `Layer`/`KmerPartition`/`KmerIndex` rewrite described above,
+   plus the `PartitionRouter` `&mut` → `&` fix (same root cause, same
+   session, do together).
+2. The future cache crate (name not chosen), consuming `KmerIndex::
+   partition(i)` — unblocks migrating `PartitionCache`/`QueryLayer` (2b).
+3. `obikalgorithm` — still deliberately waiting for a third `run()`-shaped
+   data point (`count_kmer` or `build_layers` migrating out of
+   `KmerIndex`/`PartitionRouter`) before extracting a shared trait; two
+   examples were judged not enough to be sure of the shape (`Fn+Sync` vs
+   `FnMut` callback bound already diverged between the two that exist).
+
 ## The problem
 
 Reading a layer's data (MPHF + matrix) is not free: `MphfLayer::open` mmaps

src/Cargo.lock

@@ -1515,6 +1515,24 @@ dependencies = [
  "xxhash-rust",
 ]
 
+[[package]]
+name = "obikderep"
+version = "0.1.0"
+dependencies = [
+ "niffler",
+ "obikindex",
+ "obikrope",
+ "obikseq",
+ "obilayeredmap",
+ "obiskbuilder",
+ "obiskio",
+ "obisys",
+ "rayon",
+ "sysinfo",
+ "tempfile",
+ "tracing",
+]
+
 [[package]]
 name = "obikentropy"
 version = "0.1.0"
@@ -1565,6 +1583,7 @@ dependencies = [
  "kodama",
  "obidebruinj",
  "obifastwrite",
+ "obikderep",
  "obikindex",
  "obikpartitionner",
  "obikphylo",
@@ -1607,6 +1626,7 @@ dependencies = [
  "memmap2",
  "niffler",
  "obicompactvec",
+ "obikderep",
  "obikindex",
  "obikrope",
  "obikseq",
@@ -1632,6 +1652,7 @@ dependencies = [
  "memmap2",
  "ndarray",
  "obicompactvec",
+ "obikderep",
  "obikindex",
  "obikpartitionner",
  "obikseq",

src/Cargo.toml

@@ -1,5 +1,5 @@
 [workspace]
 resolver = "3"
-members = ["obikseq", "obiread", "obiskbuilder", "obifastwrite", "obikmer","obikrope","obipipeline", "obikpartitionner","obiskio","obidebruinj","obilayeredmap", "obicompactvec", "obisys", "obikindex", "obitaxonomy", "obikentropy", "obikphylo", "obikpartition"]
+members = ["obikseq", "obiread", "obiskbuilder", "obifastwrite", "obikmer","obikrope","obipipeline", "obikpartitionner","obiskio","obidebruinj","obilayeredmap", "obicompactvec", "obisys", "obikindex", "obitaxonomy", "obikentropy", "obikphylo", "obikpartition", "obikderep"]
 [profile.release]
 debug = 1

src/obikderep/Cargo.toml

@@ -0,0 +1,21 @@
+[package]
+name = "obikderep"
+version = "0.1.0"
+edition = "2024"
+
+[dependencies]
+niffler = "3.0.0"
+obikseq       = { path = "../obikseq" }
+obikindex     = { path = "../obikindex" }
+obilayeredmap = { path = "../obilayeredmap" }
+obiskio       = { path = "../obiskio" }
+obisys        = { path = "../obisys" }
+rayon = "1"
+sysinfo = "0.39"
+tracing = "0.1.44"
+
+[dev-dependencies]
+tempfile = "3"
+obikseq  = { path = "../obikseq", features = ["test-utils"] }
+obikrope = { path = "../obikrope" }
+obiskbuilder = { path = "../obiskbuilder" }

src/obikderep/src/dereplicate.rs

@@ -1,3 +1,7 @@
+//! Per-partition dereplication mechanics — private to this crate.
+//! [`crate::Dereplicator`] is the public entry point; this module is the
+//! two-phase split+merge algorithm it runs once per partition.
+
 use std::collections::HashMap;
 use std::fs;
 use std::io;
@@ -5,16 +9,21 @@ use std::path::{Path, PathBuf};
 
 use tracing::debug;
 
-use niffler::Level;
 use niffler::send::compression::Format;
-use obikseq::Sequence;
+use niffler::Level;
 use obikseq::superkmer::SuperKmer;
+use obikseq::Sequence;
+use obilayeredmap::{dereplicated_superkmers_path, raw_superkmers_path};
 use obiskio::{SKFileMeta, SKFileReader, SKFileWriter, SKResult};
 
-use super::SK_EXT;
+/// Scratch-file extension for this algorithm's own intermediate split
+/// buckets — never read by anything outside [`dereplicate_partition`],
+/// unlike `raw`/`dereplicated` (see `obilayeredmap::{raw_superkmers_path,
+/// dereplicated_superkmers_path}`, the actual cross-crate contract).
+const TEMP_EXT: &str = "skmer.zst";
 
-/// Estimate the number of in-memory buckets needed to deduplicate the partition
+/// Estimate the number of in-memory buckets needed to deduplicate the
-/// file at `raw_path` given `available_bytes` of free RAM.
+/// partition file at `raw_path` given `available_bytes` of free RAM.
 ///
 /// Memory per HashMap entry:
 ///   key Box (1 + avg_seq_bytes) + SuperKmer header (4 B) + avg seq bytes + u64 count (8 B),
@@ -22,7 +31,7 @@ use super::SK_EXT;
 ///
 /// Returns 1 if the partition fits comfortably in memory (no split needed).
 /// Always returns a power of two.
-pub(super) fn optimal_buckets(raw_path: &Path, available_bytes: u64) -> usize {
+pub(crate) fn optimal_buckets(raw_path: &Path, available_bytes: u64) -> usize {
     // Use 60 % of available RAM to leave headroom for the rest of the process.
     let budget = (available_bytes as f64 * 0.60) as u64;
 
@@ -62,14 +71,15 @@ fn remove_skmer_file(path: &Path) -> SKResult<()> {
 /// Maximum value that fits in the 24-bit COUNT field of a SuperKmer header.
 const MAX_SK_COUNT: u64 = (1 << 24) - 1;
 
-/// Deduplicate one partition directory in place (two-phase split + merge).
+/// Deduplicate one partition's layer-0 directory in place (two-phase split
-pub(super) fn dereplicate_partition(dir: &Path, level: Level, n_temp: usize) -> SKResult<()> {
+/// + merge): `raw_superkmers_path(dir)` -> `dereplicated_superkmers_path(dir)`.
-    let raw_path = dir.join(format!("raw.{SK_EXT}"));
+pub(crate) fn dereplicate_partition(dir: &Path, level: Level, n_temp: usize) -> SKResult<()> {
+    let raw_path = raw_superkmers_path(dir);
     if !raw_path.exists() {
         return Ok(());
     }
 
-    let out_path = dir.join(format!("dereplicated.{SK_EXT}"));
+    let out_path = dereplicated_superkmers_path(dir);
     let mut writer = SKFileWriter::create_with(&out_path, Format::Zstd, level)?;
 
     if n_temp == 1 {
@@ -81,7 +91,7 @@ pub(super) fn dereplicate_partition(dir: &Path, level: Level, n_temp: usize) ->
         // ── Phase 1: split raw file into temp buckets ─────────────────────────
         let temp_mask = (n_temp as u64) - 1;
         let temp_paths: Vec<PathBuf> = (0..n_temp)
-            .map(|j| dir.join(format!("temp_{j:04}.{SK_EXT}")))
+            .map(|j| dir.join(format!("temp_{j:04}.{TEMP_EXT}")))
             .collect();
 
         {

src/obikderep/src/lib.rs

@@ -0,0 +1,110 @@
+//! Superkmer dereplication — the second stage of the indexing pipeline,
+//! after `obikpartitionner::PartitionRouter::run` (scatter) has written
+//! each partition's raw superkmer file, before
+//! `obikpartitionner::PartitionRouter::count_kmer` (counting) reads the
+//! result. One algorithm, one crate — see
+//! `DevDocMD/implementation/partition_layer_cache.md`'s "on avance pas à
+//! pas" note: `obikpartitionner` used to also own dereplication and
+//! counting; this crate is step one of splitting that bundle apart,
+//! deliberately one algorithm at a time rather than all at once, so a
+//! shared `Algorithm` pattern can be factored out later from real
+//! examples instead of guessed at up front.
+
+mod dereplicate;
+
+use std::sync::atomic::{AtomicU64, Ordering};
+
+use niffler::Level;
+use obikindex::KmerIndex;
+use obiskio::SKResult;
+use obisys::Progress;
+use rayon::prelude::*;
+use sysinfo::System;
+
+use dereplicate::{dereplicate_partition, optimal_buckets};
+
+/// Deduplicates every partition's raw superkmer file in place, replacing
+/// each with a dereplicated one where identical canonical sequences are
+/// merged and their counts summed.
+///
+/// Two-phase construction, same shape as `PartitionRouter`: `new` (no disk
+/// access), then `run`. Nothing to configure yet — no setters, unlike
+/// `PartitionRouter`'s `level_max`/`theta`/etc. — added if a real need
+/// shows up, not speculatively.
+pub struct Dereplicator<'a> {
+    index: &'a KmerIndex,
+    n_partitions: usize,
+    level: Level,
+}
+
+impl<'a> Dereplicator<'a> {
+    pub fn new(index: &'a KmerIndex) -> Self {
+        Self {
+            index,
+            n_partitions: index.n_partitions(),
+            level: Level::One,
+        }
+    }
+
+    /// Dereplicate every partition in parallel.
+    ///
+    /// Each partition file is processed in two phases to bound memory use:
+    ///
+    /// 1. **Split** — the raw file is scattered into `2^temp_bits` temporary
+    ///    files routed by `hash(canonical_seq) & temp_mask`. Because duplicates
+    ///    always share the same hash, they always land in the same temp file.
+    /// 2. **Merge** — each temp file is loaded fully into a `HashMap`, counts
+    ///    are accumulated in `u64` (no 24-bit overflow risk), and the result is
+    ///    appended to the partition's dereplicated file.
+    ///
+    /// If a merged count exceeds the 24-bit header limit, the sequence is
+    /// emitted as multiple records whose counts sum to the true total.
+    ///
+    /// `on_progress`, when set, is called once per completed partition, from
+    /// whichever rayon worker thread finished it — `Fn(...) + Sync`, not
+    /// `FnMut`, unlike `PartitionRouter::run`'s callback: that one is driven
+    /// from a single sequential loop, this one from a parallel `par_iter`,
+    /// so the callback itself must tolerate concurrent calls (same reason
+    /// `obisys::TracedBar`'s own methods take `&self`, not `&mut self`).
+    /// `total: Some(n_partitions)` — known up front here, unlike
+    /// `PartitionRouter::run`'s bases-processed count, so the caller can
+    /// render an actual progress bar rather than a spinner. This crate
+    /// never renders anything itself — see the module docs.
+    pub fn run(&self, on_progress: Option<impl Fn(Progress) + Sync>) -> SKResult<()> {
+        let level = self.level;
+        let sys = System::new_all();
+        // available_memory() can return 0 on macOS when the compressor page count exceeds
+        // free+inactive+purgeable pages (sysinfo saturating_sub). Fall back to half of total.
+        let available = match sys.available_memory() {
+            0 => sys.total_memory() / 2,
+            n => n,
+        };
+        let n_threads = rayon::current_num_threads().max(1) as u64;
+        let available_per_thread = available / n_threads;
+        let done = AtomicU64::new(0);
+
+        let results: Vec<SKResult<()>> = (0..self.n_partitions)
+            .into_par_iter()
+            .map(|i| {
+                let dir = self.index.layer_dir(i, 0);
+                let result = if dir.exists() {
+                    let raw_path = obilayeredmap::raw_superkmers_path(&dir);
+                    let n_buckets = optimal_buckets(&raw_path, available_per_thread);
+                    dereplicate_partition(&dir, level, n_buckets)
+                } else {
+                    Ok(())
+                };
+                if let Some(cb) = &on_progress {
+                    let pos = done.fetch_add(1, Ordering::Relaxed) + 1;
+                    cb(Progress { position: pos, total: Some(self.n_partitions as u64) });
+                }
+                result
+            })
+            .collect();
+
+        for r in results {
+            r?;
+        }
+        Ok(())
+    }
+}

src/obikindex/src/index_layer.rs

@@ -43,7 +43,7 @@ impl KmerIndex {
         block_bits: u8,
     ) -> Result<usize, SKError> {
         let layer0_dir = self.layer_dir(i, 0);
-        let dedup_path = layer0_dir.join("dereplicated.skmer.zst");
+        let dedup_path = obilayeredmap::dereplicated_superkmers_path(&layer0_dir);
         if !dedup_path.exists() {
             return Ok(0);
         }
@@ -123,7 +123,7 @@ impl KmerIndex {
     /// Deletes `dereplicated.skmer.zst` (+ sidecar), `mphf1.bin`, `counts1.bin`.
     pub fn remove_build_artifacts(&self, i: usize) {
         let layer0_dir = self.layer_dir(i, 0);
-        let dedup = layer0_dir.join("dereplicated.skmer.zst");
+        let dedup = obilayeredmap::dereplicated_superkmers_path(&layer0_dir);
         remove_if_exists(&SKFileMeta::sidecar_path(&dedup));
         remove_if_exists(&dedup);
         remove_if_exists(&layer0_dir.join("mphf1.bin"));

src/obikmer/Cargo.toml

@@ -16,6 +16,7 @@ obidebruinj   = { path = "../obidebruinj" }
 obipipeline   = { path = "../obipipeline" }
 obikrope      = { path = "../obikrope" }
 obikpartitionner = { path = "../obikpartitionner" }
+obikderep     = { path = "../obikderep" }
 obisys        = { path = "../obisys" }
 obiskio       = { path = "../obiskio" }
 obikindex     = { path = "../obikindex", default-features = false }

src/obikmer/src/cmd/index/mod.rs

@@ -2,6 +2,7 @@ use std::path::PathBuf;
 use std::time::Instant;
 
 use clap::Args;
+use obikderep::Dereplicator;
 use obikindex::{validate_label, GenomeInfo, IndexConfig, IndexState, KmerIndex};
 use obikpartitionner::PartitionRouter;
 use obilayeredmap::IndexMode;
@@ -10,7 +11,7 @@ fn parse_key_value(s: &str) -> Result<(String, String), String> {
     let pos = s.find('=').ok_or_else(|| format!("invalid key=value: no '=' in '{s}'"))?;
     Ok((s[..pos].to_string(), s[pos + 1..].to_string()))
 }
-use obisys::{spinner, Progress, Reporter, Stage};
+use obisys::{progress_bar, spinner, Progress, Reporter, Stage};
 use tracing::info;
 
 use crate::cli::{CommonArgs, block_size_to_bits, partitions_to_bits};
@@ -290,15 +291,19 @@ pub fn run(args: IndexArgs) {
 
     // ── Stage 2: dereplicate + count ─────────────────────────────────────────
     if idx.state() < IndexState::Counted {
-        let router = PartitionRouter::new(&mut idx);
-
         let t = Stage::start("dereplicate");
-        router.dereplicate().unwrap_or_else(|e| {
+        let pb = progress_bar("dereplication", idx.n_partitions() as u64, "partitions");
+        Dereplicator::new(&idx)
+            .run(Some(|_: Progress| pb.inc(1)))
+            .unwrap_or_else(|e| {
                 eprintln!("error: {e}");
                 std::process::exit(1);
             });
+        pb.finish_and_clear();
         rep.push(t.stop());
 
+        let router = PartitionRouter::new(&mut idx);
+
         let t = Stage::start("count_kmer");
         let spectrum = router.count_kmer(args.keep_intermediate).unwrap_or_else(|e| {
             eprintln!("error: {e}");

src/obikpartitionner/Cargo.toml

@@ -7,6 +7,7 @@ edition = "2024"
 tempfile = "3"
 obikseq   = { path = "../obikseq",   features = ["test-utils"] }
 obikrope  = { path = "../obikrope" }
+obikderep = { path = "../obikderep" }
 
 [dependencies]
 niffler = "3.0.0"

src/obikpartitionner/src/partition/mod.rs

@@ -1,20 +1,18 @@
 //! K-mer partitioning: routing super-kmers into per-partition, layer-0
-//! files, deduplicating them, and counting unique canonical k-mers.
+//! files, and counting unique canonical k-mers. Dereplication itself moved
+//! to `obikderep` (2026-08-20) — see
+//! `DevDocMD/implementation/partition_layer_cache.md`: one algorithm, one
+//! crate, split off one at a time rather than all at once.
 //!
 //! Submodules: [`router`] (`PartitionRouter`, `KmerSpectrum`, the
-//! routing/lifecycle API — partition/layer path naming itself lives on
+//! routing/counting lifecycle API — partition/layer path naming itself
-//! `obikindex::KmerIndex`, not here, see
+//! lives on `obikindex::KmerIndex`, not here), [`count`] (unique-kmer
-//! `DevDocMD/implementation/partition_layer_cache.md`), [`dereplicate`]
-//! (two-phase split+merge deduplication), [`count`] (unique-kmer
 //! enumeration, MPHF, abundance counting).
 
 mod count;
-mod dereplicate;
 mod router;
 
 #[cfg(test)]
 mod tests;
 
 pub use router::{KmerSpectrum, PartitionRouter};
-
-const SK_EXT: &str = "skmer.zst";

src/obikpartitionner/src/partition/router.rs

@@ -25,8 +25,6 @@ use obiread::NucPage;
 use crate::kmer_sort::chunk_size_from_ram;
 
 use super::count::count_partition;
-use super::dereplicate::{dereplicate_partition, optimal_buckets};
-use super::SK_EXT;
 
 pub struct KmerSpectrum {
     pub f0: u64,
@@ -269,60 +267,6 @@ impl<'a> PartitionRouter<'a> {
         self.close()
     }
 
-    /// Deduplicate all `raw.{ext}` files in parallel, replacing each with a
-    /// `dereplicated.{ext}` file where identical canonical sequences are merged
-    /// and their counts summed.
-    ///
-    /// Each partition file is processed in two phases to bound memory use:
-    ///
-    /// 1. **Split** — the raw file is scattered into `2^temp_bits` temporary
-    ///    files routed by `hash(canonical_seq) & temp_mask`.  Because duplicates
-    ///    always share the same hash, they always land in the same temp file.
-    /// 2. **Merge** — each temp file is loaded fully into a `HashMap`, counts
-    ///    are accumulated in `u64` (no 24-bit overflow risk), and the result is
-    ///    appended to `dereplicated.{ext}`.
-    ///
-    /// If a merged count exceeds the 24-bit header limit, the sequence is
-    /// emitted as multiple records whose counts sum to the true total.
-    pub fn dereplicate(&self) -> SKResult<()> {
-        let level = self.level;
-        let sys = System::new_all();
-        // available_memory() can return 0 on macOS when the compressor page count exceeds
-        // free+inactive+purgeable pages (sysinfo saturating_sub). Fall back to half of total.
-        let available = match sys.available_memory() {
-            0 => sys.total_memory() / 2,
-            n => n,
-        };
-        let n_threads = rayon::current_num_threads().max(1) as u64;
-        let available_per_thread = available / n_threads;
-
-        let pb = progress_bar("dereplication", self.n_partitions as u64, "partitions");
-
-        let results: Vec<SKResult<()>> = (0..self.n_partitions)
-            .into_par_iter()
-            .map(|i| {
-                let dir = self.layer0_dir(i);
-                if !dir.exists() {
-                    pb.inc(1);
-                    return Ok(());
-                }
-                let raw_path = dir.join(format!("raw.{SK_EXT}"));
-                let t = Instant::now();
-                let n_buckets = optimal_buckets(&raw_path, available_per_thread);
-                let result = dereplicate_partition(&dir, level, n_buckets);
-                pb.set_message(format!("last {:.0}ms", t.elapsed().as_millis()));
-                pb.inc(1);
-                result
-            })
-            .collect();
-
-        pb.finish_and_clear();
-        for r in results {
-            r?;
-        }
-        Ok(())
-    }
-
     /// For each partition that has a `dereplicated.{ext}` file:
     /// 1. Enumerates all unique canonical kmers (two passes over the file).
     /// 2. Builds a provisional MPHF (FMPHGO) over those kmers.
@@ -347,7 +291,7 @@ impl<'a> PartitionRouter<'a> {
             .into_par_iter()
             .map(|i| {
                 let dir = self.layer0_dir(i);
-                let dedup_path = dir.join(format!("dereplicated.{SK_EXT}"));
+                let dedup_path = obilayeredmap::dereplicated_superkmers_path(&dir);
                 if !dedup_path.exists() {
                     pb.inc(1);
                     return Ok(());
@@ -416,7 +360,7 @@ impl<'a> PartitionRouter<'a> {
         if self.writers[partition].is_none() {
             let dir = self.layer0_dir(partition);
             Layer::create(&dir).map_err(|e| io::Error::other(e.to_string()))?;
-            let file_path = dir.join(format!("raw.{SK_EXT}"));
+            let file_path = obilayeredmap::raw_superkmers_path(&dir);
             let writer = SKFileWriter::create_with(file_path, Format::Zstd, self.level)?;
             self.writers[partition] = Some(writer);
         }

src/obikpartitionner/src/partition/tests.rs

@@ -1,6 +1,7 @@
 use std::collections::HashMap;
 use std::fs;
 
+use obikderep::Dereplicator;
 use obikindex::{IndexConfig, KmerIndex};
 use obikrope::Rope;
 use obikseq::SuperKmer;
@@ -64,11 +65,11 @@ fn pipeline_counts(seqs: &[&[u8]]) -> (u64, u64) {
     let mut kp = PartitionRouter::new(&mut index);
     kp.write_batch(superkmers).unwrap();
     kp.close().unwrap();
-    kp.dereplicate().unwrap();
     drop(kp); // ends the borrow of `index` early — `PartitionRouter`'s `Drop` impl would otherwise extend it to the end of scope
+    Dereplicator::new(&index).run(None::<fn(obisys::Progress)>).unwrap();
 
     let part_dir = index.layer_dir(0, 0);
-    let dedup_path = part_dir.join("dereplicated.skmer.zst");
+    let dedup_path = obilayeredmap::dereplicated_superkmers_path(&part_dir);
     if !dedup_path.exists() {
         return (0, 0);
     }

src/obikphylo/Cargo.toml

@@ -21,5 +21,6 @@ tracing          = "0.1.44"
 
 [dev-dependencies]
 obiread   = { path = "../obiread" }
+obikderep = { path = "../obikderep" }
 tempfile = "3"
 tracing-subscriber = { version = "0.3", features = ["fmt", "env-filter"] }

src/obikphylo/src/siblings/tests.rs

@@ -6,6 +6,7 @@ use obilayeredmap::MphfLayer;
 use obisys::Reporter;
 use tempfile::tempdir;
 
+use obikderep::Dereplicator;
 use obikindex::{GenomeInfo, IndexConfig, KmerIndex, MergeMode};
 use obikpartitionner::PartitionRouter;
 
@@ -76,7 +77,10 @@ fn build_single_genome_index(dir: &Path, label: &str, seq: &[u8]) -> KmerIndex {
         router.write_batch(batch).expect("write_batch");
     }
     router.close().expect("close partition writers");
-    router.dereplicate().expect("dereplicate");
+    drop(router); // ends the borrow of `idx` early — `PartitionRouter`'s `Drop` impl would otherwise extend it to the end of scope
+
+    Dereplicator::new(&idx).run(None::<fn(obisys::Progress)>).expect("dereplicate");
+    let router = PartitionRouter::new(&mut idx);
     let spectrum = router.count_kmer(false).expect("count_kmer");
     drop(router); // ends the borrow of `idx` early — `PartitionRouter`'s `Drop` impl would otherwise extend it to the end of scope
 

src/obilayeredmap/src/layer.rs

@@ -33,13 +33,37 @@ pub trait LayerData: Sized {
 ///
 /// `obilayeredmap` operates within a single partition's index root; it has
 /// no notion of "partition" at all. Turning a partition number into that
-/// root is `obikpartitionner::KmerPartition::part_dir`'s job, one layer up
+/// root is `obikindex::KmerIndex::index_dir`'s job, one layer up — callers
-/// — callers here only ever name a layer *number*, never build the path
+/// here only ever name a layer *number*, never build the path themselves.
-/// themselves.
 pub fn layer_dir(root: &Path, i: usize) -> PathBuf {
     root.join(format!("layer_{i}"))
 }
 
+/// Superkmer-file extension used by every pre-layer-construction artifact
+/// below (`raw`/`dereplicated`) — an implementation detail of the SK file
+/// format, never meant to leak as a literal string past this module.
+const SK_EXT: &str = "skmer.zst";
+
+/// Path of a layer's raw, not-yet-dereplicated superkmer file — written by
+/// whichever algorithm routes superkmers into this layer (today:
+/// `obikpartitionner::PartitionRouter`), read by whichever algorithm
+/// dereplicates it (today: `obikderep::Dereplicator`). Naming this once
+/// here, rather than in either algorithm crate, is what lets two
+/// independent crates agree on the filename without depending on each
+/// other — see `DevDocMD/implementation/partition_layer_cache.md`.
+pub fn raw_superkmers_path(layer_dir: &Path) -> PathBuf {
+    layer_dir.join(format!("raw.{SK_EXT}"))
+}
+
+/// Path of a layer's dereplicated superkmer file — written by
+/// `obikderep::Dereplicator`, read by whichever algorithm counts kmer
+/// abundances from it (today: `obikpartitionner::PartitionRouter::
+/// count_kmer`) and, later, by `obikindex::build_index_layer` to build the
+/// real layer.
+pub fn dereplicated_superkmers_path(layer_dir: &Path) -> PathBuf {
+    layer_dir.join(format!("dereplicated.{SK_EXT}"))
+}
+
 /// Opens layer `i`'s data only, skipping the MPHF — for callers that only
 /// need matrix-level operations (distance traits, column weights, group
 /// filters, sub-matrix extraction) and never look up a kmer for this layer.

src/obilayeredmap/src/lib.rs

@@ -10,7 +10,10 @@ pub(crate) mod mphf_layer;
 
 pub use content_layer::Layer;
 pub use error::{OLMError, OLMResult};
-pub use layer::{layer_dir, open_data, HasLayerContent, HasStorageKind, Hit, LayerContent, LayerData, TypedLayer};
+pub use layer::{
+    dereplicated_superkmers_path, layer_dir, open_data, raw_superkmers_path, HasLayerContent,
+    HasStorageKind, Hit, LayerContent, LayerData, TypedLayer,
+};
 pub use layered_store::LayeredStore;
 pub use map::LayeredMap;
 pub use meta::{IndexMode, PartitionMeta};