MRG: update to code for forthcoming sourmash release

.github/workflows/build-test.yml

@@ -29,7 +29,6 @@ jobs:
         auto-update-conda: true
         python-version: 3.12
         channels: conda-forge,bioconda
-        miniforge-variant: Mambaforge
         miniforge-version: latest
         use-mamba: true
         mamba-version: "*"

Cargo.toml

@@ -12,7 +12,7 @@ crate-type = ["cdylib"]
 pyo3 = { version = "0.22.4", features = ["extension-module", "anyhow"] }
 rayon = "1.10.0"
 serde = { version = "1.0.210", features = ["derive"] }
-sourmash = { version = "0.15.2", features = ["branchwater"] }
+sourmash = { version = "0.16.0", features = ["branchwater"] }
 serde_json = "1.0.128"
 niffler = "2.4.0"
 log = "0.4.22"

src/manysearch.rs

@@ -219,6 +219,7 @@ fn downsample_and_inflate_abundances(
     // avoid downsampling if we can
     if against_scaled != query_scaled {
         let against_ds = against
+            .clone()
             .downsample_scaled(query.scaled())
             .expect("cannot downsample sketch");
         (abunds, sum_weighted) = query.inflated_abundances(&against_ds)?;

src/python/tests/test_fastmultigather.py

@@ -959,15 +959,15 @@ def test_indexed_full_output(runtmp):
     # check a few columns
     average_ani = set(df['average_containment_ani'])
     avg_ani = set([round(x, 4) for x in average_ani])
-    assert avg_ani == {0.8502, 0.8584, 0.8602}
+    assert avg_ani == {0.9221, 0.9306, 0.9316} # @CTB check against py gather
 
     f_unique_weighted = set(df['f_unique_weighted'])
     f_unique_weighted = set([round(x, 4) for x in f_unique_weighted])
     assert f_unique_weighted == {0.0063, 0.002, 0.0062}
 
     unique_intersect_bp = set(df['unique_intersect_bp'])
     unique_intersect_bp = set([round(x,4) for x in unique_intersect_bp])
-    assert unique_intersect_bp == {44000, 18000, 22000}
+    assert unique_intersect_bp == {4400000, 1800000, 2200000}
 
 
 def test_nonindexed_full_vs_sourmash_gather(runtmp):

src/utils.rs

@@ -476,25 +476,30 @@ pub fn load_sketches_above_threshold(
             let mut results = Vec::new();
             // Load against into memory
             if let Ok(against_sig) = against_collection.sig_from_record(against_record) {
-                if let Some(against_mh) = against_sig.minhash() {
-                    // downsample against_mh, but keep original md5sum
-                    let against_mh_ds = against_mh.downsample_scaled(query.scaled()).unwrap();
-                    if let Ok(overlap) = against_mh_ds.count_common(query, false) {
-                        if overlap >= threshold_hashes {
-                            let result = PrefetchResult {
-                                name: against_record.name().to_string(),
-                                md5sum: against_mh.md5sum(),
-                                minhash: against_mh_ds.clone(),
-                                location: against_record.internal_location().to_string(),
-                                overlap,
-                            };
-                            results.push(result);
-                        }
+                let against_filename = against_sig.filename();
+                let against_mh: KmerMinHash = against_sig.try_into().expect("cannot get sketch");
+                let against_md5 = against_mh.md5sum(); // keep original md5sum
+
+                let against_mh_ds = against_mh
+                    .downsample_scaled(query.scaled())
+                    .expect("cannot downsample sketch");
+
+                // good? ok, store as candidate from prefetch.
+                if let Ok(overlap) = against_mh_ds.count_common(query, false) {
+                    if overlap >= threshold_hashes {
+                        let result = PrefetchResult {
+                            name: against_record.name().to_string(),
+                            md5sum: against_md5,
+                            minhash: against_mh_ds,
+                            location: against_record.internal_location().to_string(),
+                            overlap,
+                        };
+                        results.push(result);
                     }
                 } else {
                     eprintln!(
                         "WARNING: no compatible sketches in path '{}'",
-                        against_sig.filename()
+                        against_filename
                     );
                     let _i = skipped_paths.fetch_add(1, atomic::Ordering::SeqCst);
                 }