Fix skani shuffle

src/clusterer.rs

@@ -635,4 +635,30 @@ mod tests {
         }
         assert_eq!(vec![vec![0, 1, 3], vec![2]], clusters)
     }
+
+    #[test]
+    fn test_skani_skani_two_preclusters() {
+        init();
+        let mut clusters = cluster(
+            &[
+                "tests/data/abisko4/73.20120800_S1X.13.fna",
+                "tests/data/abisko4/73.20120600_S2D.19.fna",
+                "tests/data/abisko4/73.20120700_S3X.12.fna",
+                "tests/data/abisko4/73.20110800_S2D.13.fna",
+                "tests/data/antonio_mags/BE_RX_R2_MAG52.fna",
+            ],
+            &crate::skani::SkaniPreclusterer {
+                threshold: 90.0,
+                min_aligned_threshold: 0.2,
+            },
+            &crate::skani::SkaniClusterer {
+                threshold: 99.0,
+                min_aligned_threshold: 0.2,
+            },
+        );
+        for cluster in clusters.iter_mut() {
+            cluster.sort_unstable();
+        }
+        assert_eq!(vec![vec![4], vec![0, 1, 3], vec![2]], clusters)
+    }
 }

src/skani.rs

@@ -42,6 +42,7 @@ fn precluster_skani(
         .iter()
         .map(|x| x.to_string())
         .collect::<Vec<String>>();
+    // Note that sketches is now shuffled!
     let sketches = &file_io::fastx_to_sketches(&fasta_strings, &sketch_params, true);
 
     // Right now implemented by parallel collection into a queue, and then
@@ -73,9 +74,21 @@ fn precluster_skani(
                 let map_params = chain::map_params_from_sketch(ref_sketch, false, &command_params);
                 let ani_result = chain::chain_seeds(ref_sketch, &sketches[j], map_params);
                 let ani = ani_result.ani * 100.;
+                let ref_name = ref_sketch.file_name.clone();
+                let query_name = sketches[j].file_name.clone();
+
+                debug!("Pushing ANI result for {} and {}", ref_name, query_name);
+                let ref_index = genome_fasta_paths
+                    .iter()
+                    .position(|&r| r == ref_name)
+                    .unwrap();
+                let query_index = genome_fasta_paths
+                    .iter()
+                    .position(|&r| r == query_name)
+                    .unwrap();
                 if ani >= threshold {
                     queue
-                        .push((i, j, ani))
+                        .push((ref_index, query_index, ani))
                         .expect("Failed to push to queue during preclustering");
                 }
             }