remove dependency on yanked quickersort

Cargo.toml

@@ -69,7 +69,6 @@ num_cpus = "1.13.1"
 bio-types = "0.12.1"
 itertools = "0.10.3"
 thiserror = "1.0.30"
-quickersort = "3.0.1"
 statrs = "0.15.0"
 rust-htslib = { git = "https://github.com/rust-bio/rust-htslib", version = "0.38.2", default-features = false, features = [
   "bzip2",

libradicl/Cargo.toml

@@ -37,7 +37,6 @@ csv = "1.1.6"
 dashmap = "^5.1.0"
 crossbeam-channel = "0.5.4"
 bio-types = "0.12.1"
-quickersort = "3.0.1"
 needletail = "0.4.1"
 flate2 = "1.0.22"
 smallvec = "1.8.0"

libradicl/src/rad_types.rs

@@ -290,7 +290,7 @@ impl ReadRecord {
         }
 
         // make sure these are sorted in this step.
-        quickersort::sort(&mut rec.refs[..]);
+        rec.refs.sort_unstable();
         rec
     }
 
@@ -332,7 +332,7 @@ impl ReadRecord {
         }
 
         // make sure these are sorted in this step.
-        quickersort::sort(&mut rec.refs[..]);
+        rec.refs.sort_unstable();
         rec
     }
 }

src/collate.rs

@@ -117,7 +117,7 @@ pub fn collate(
 
     // sort this so that we deal with largest cells (by # of reads) first
     // sort in _descending_ order by count.
-    quickersort::sort_by_key(&mut tsv_map[..], |&a: &(u64, u64)| std::cmp::Reverse(a.1));
+    tsv_map.sort_unstable_by_key(|&a: &(u64, u64)| std::cmp::Reverse(a.1));
 
     /*
     let est_num_rounds = (total_to_collate as f64 / max_records as f64).ceil() as u64;

src/eq_class.rs

@@ -283,7 +283,7 @@ impl EqMap {
             })
             .collect();
 
-        quickersort::sort(&mut hash_vec[..]);
+        hash_vec.sort_unstable();
 
         let mut prev_hash = 0u64;
         let mut prev_umi = u64::MAX;
@@ -408,7 +408,7 @@ impl EqMap {
 
             let v = &mut self.eqc_info[idx];
             // sort so dups are adjacent
-            quickersort::sort(&mut v.umis[..]);
+            v.umis.sort_unstable();
             // we need a copy of the vector b/c we
             // can't easily modify it in place
             // at least I haven't seen how (@k3yavi, help here if you can).

src/pugutils.rs

@@ -390,7 +390,7 @@ fn resolve_num_molecules_crlike_from_vec_prefer_ambig(
     // first on umi
     // then on gene_id
     // then on count
-    quickersort::sort(umi_gene_count_vec);
+    umi_gene_count_vec.sort_unstable();
 
     // hold the current umi and gene we are examining
     let mut curr_umi = umi_gene_count_vec.first().expect("cell with no UMIs").0;
@@ -524,7 +524,7 @@ fn resolve_num_molecules_crlike_from_vec(
     // first on umi
     // then on gene_id
     // then on count
-    quickersort::sort(umi_gene_count_vec);
+    umi_gene_count_vec.sort_unstable();
 
     // hold the current umi and gene we are examining
     let mut curr_umi = umi_gene_count_vec.first().expect("cell with no UMIs").0;
@@ -640,7 +640,7 @@ pub fn get_num_molecules_cell_ranger_like_small(
             .map(|tid| tid_to_gid[*tid as usize])
             .collect();
         // and make the gene ids unique
-        quickersort::sort(&mut gset[..]);
+        gset.sort_unstable();
         gset.dedup();
         for g in &gset {
             umi_gene_count_vec.push((umi, *g, 1));
@@ -693,7 +693,7 @@ pub fn get_num_molecules_cell_ranger_like(
         // necessarily be adjacent here, since
         // they are always asigned adjacent ids
         // with spliced being even and unspliced odd.
-        quickersort::sort(&mut gset[..]);
+        gset.sort_unstable();
         gset.dedup();
 
         // add every (umi, count), gene pair as a triplet
@@ -772,7 +772,7 @@ pub fn get_num_molecules_trivial_discard_all_ambig(
     // equivalence classes that still map to the same
     // gene.
     for (k, v) in gene_map.iter_mut() {
-        quickersort::sort(&mut v[..]);
+        v.sort_unstable();
         v.dedup();
         // the count is the number of distinct UMIs.
         counts[*k as usize] += v.len() as f32;
@@ -827,7 +827,7 @@ fn get_num_molecules_large_component(
             .map(|tid| tid_to_gid[*tid as usize])
             .collect();
         // and make the gene ids unique
-        quickersort::sort(&mut gset[..]);
+        gset.sort_unstable();
         gset.dedup();
 
         // add every (umi, count), gene pair as a triplet
@@ -843,7 +843,7 @@ fn get_num_molecules_large_component(
     // first on umi
     // then on gene_id
     // then on count
-    quickersort::sort(&mut umi_gene_count_vec[..]);
+    umi_gene_count_vec.sort_unstable();
 
     // hold the current umi and gene we are examining
     let mut curr_umi = umi_gene_count_vec.first().expect("cell with no UMIs").0;
@@ -1110,7 +1110,7 @@ pub fn get_num_molecules(
                     .map(|i| tid_to_gid[i as usize])
                     .collect();
                 // sort since we will be hashing the ordered vector
-                quickersort::sort(&mut global_genes[..]);
+                global_genes.sort_unstable();
                 // dedup as well since we don't care about duplicates
                 global_genes.dedup();
 
@@ -1154,7 +1154,7 @@ pub fn get_num_molecules(
             }
 
             let mut global_genes: Vec<u32> = tl.iter().map(|i| tid_to_gid[*i as usize]).collect();
-            quickersort::sort(&mut global_genes[..]);
+            global_genes.sort_unstable();
             global_genes.dedup();
 
             // extract gene-level eqclass and increment count by 1

src/utils.rs

@@ -537,7 +537,6 @@ pub fn extract_usa_eqmap(
                 // however, because we know the original eqc labels
                 // and the USA mode labels are 1-1, we don't need this
                 // so avoid the sort here.
-                // quickersort::sort(tvec);
                 idx_eq_list.add_label_vec(tvec.as_slice());
                 eq_id_count.push((ctr as u32, *count));
             }