fix needletail in benchmarks

sourmash-bio · Jul 13, 2020 · 14c56da · 14c56da
1 parent 58bbaac
commit 14c56da
Show file tree

Hide file tree

Showing 2 changed files with 29 additions and 35 deletions.
diff --git a/src/core/benches/compute.rs b/src/core/benches/compute.rs
@@ -4,7 +4,7 @@ extern crate criterion;
 use std::fs::File;
 use std::io::{Cursor, Read};
 
-use needletail::parse_sequence_reader;
+use needletail::parse_fastx_reader;
 use sourmash::cmd::ComputeParameters;
 use sourmash::signature::Signature;
 
@@ -35,59 +35,43 @@ fn add_sequence(c: &mut Criterion) {
         b.iter(|| {
             let fasta_data = Cursor::new(data_upper.clone());
             let mut sig = template_sig.clone();
-            parse_sequence_reader(
-                fasta_data,
-                |_| {},
-                |rec| {
-                    sig.add_sequence(&rec.seq, false).unwrap();
-                },
-            )
-            .unwrap();
+            let mut parser = parse_fastx_reader(fasta_data).unwrap();
+            while let Some(rec) = parser.next() {
+                sig.add_sequence(&rec.unwrap().seq(), false).unwrap();
+            }
         });
     });
 
     group.bench_function("lowercase", |b| {
         b.iter(|| {
             let fasta_data = Cursor::new(data_lower.clone());
             let mut sig = template_sig.clone();
-            parse_sequence_reader(
-                fasta_data,
-                |_| {},
-                |rec| {
-                    sig.add_sequence(&rec.seq, false).unwrap();
-                },
-            )
-            .unwrap();
+            let mut parser = parse_fastx_reader(fasta_data).unwrap();
+            while let Some(rec) = parser.next() {
+                sig.add_sequence(&rec.unwrap().seq(), false).unwrap();
+            }
         });
     });
 
     group.bench_function("invalid kmers", |b| {
         b.iter(|| {
             let fasta_data = Cursor::new(data_errors.clone());
             let mut sig = template_sig.clone();
-            parse_sequence_reader(
-                fasta_data,
-                |_| {},
-                |rec| {
-                    sig.add_sequence(&rec.seq, true).unwrap();
-                },
-            )
-            .unwrap();
+            let mut parser = parse_fastx_reader(fasta_data).unwrap();
+            while let Some(rec) = parser.next() {
+                sig.add_sequence(&rec.unwrap().seq(), true).unwrap();
+            }
         });
     });
 
     group.bench_function("force with valid kmers", |b| {
         b.iter(|| {
             let fasta_data = Cursor::new(data_upper.clone());
             let mut sig = template_sig.clone();
-            parse_sequence_reader(
-                fasta_data,
-                |_| {},
-                |rec| {
-                    sig.add_sequence(&rec.seq, true).unwrap();
-                },
-            )
-            .unwrap();
+            let mut parser = parse_fastx_reader(fasta_data).unwrap();
+            while let Some(rec) = parser.next() {
+                sig.add_sequence(&rec.unwrap().seq(), true).unwrap();
+            }
         });
     });
 }

diff --git a/src/core/src/sketch/minhash.rs b/src/core/src/sketch/minhash.rs
@@ -592,7 +592,11 @@ impl KmerMinHash {
             first.count_common(&downsampled_mh, false)
         } else {
             self.check_compatible(other)?;
-            let iter = Intersection::new(self.mins.iter(), other.mins.iter());
+            let iter = if self.size() < other.size() {
+                Intersection::new(self.mins.iter(), other.mins.iter())
+            } else {
+                Intersection::new(other.mins.iter(), self.mins.iter())
+            };
 
             Ok(iter.count() as u64)
         }
@@ -669,6 +673,8 @@ impl KmerMinHash {
             unimplemented!() // @CTB fixme
         }
 
+        // TODO: check which one is smaller, swap around if needed
+
         let abunds = self.abunds.as_ref().unwrap();
         let other_abunds = other.abunds.as_ref().unwrap();
 
@@ -1698,7 +1704,11 @@ impl KmerMinHashBTree {
             first.count_common(&downsampled_mh, false)
         } else {
             self.check_compatible(other)?;
-            let iter = Intersection::new(self.mins.iter(), other.mins.iter());
+            let iter = if self.size() < other.size() {
+                Intersection::new(self.mins.iter(), other.mins.iter())
+            } else {
+                Intersection::new(other.mins.iter(), self.mins.iter())
+            };
 
             Ok(iter.count() as u64)
         }