We need an owned type to update a single DashMap in parallel, but we …

…can use a Box<[u8]> for the hashmap keys instead of Vec<u8>.
suchapalaver · Oct 7, 2021 · 8d6c976 · 8d6c976
1 parent 7d9ff74
commit 8d6c976
Show file tree

Hide file tree

Showing 2 changed files with 26 additions and 27 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/src/lib.rs b/src/lib.rs
@@ -37,30 +37,29 @@ pub fn run(config: Config) -> Result<(), Box<dyn Error>> {
     let k: usize = config.kmer_len;
 
     // 'N'
-    let n: u8 = 78;
+    const N: u8 = b'N';
 
     //  Create fasta file reader
     let reader: fasta::Reader<std::io::BufReader<File>> =
         fasta::Reader::from_file(&filepath).unwrap();
 
     //  Create a DashMap
-    let fasta_hash: DashMap<Vec<u8>, u32> = DashMap::new();
+    let fasta_hash: DashMap<Box<[u8]>, u32> = DashMap::new();
 
     //  Read fasta records into a Dashmap, a hashmap mutably accessible from different parallel processes
     reader
         .records()
         .into_iter()
         .par_bridge()
-        .for_each(|result| {
-            let seq: &[u8] = result.as_ref().unwrap().seq();
+        .for_each(|record| {
+            let seq: &[u8] = record.as_ref().unwrap().seq();
 
             for i in 0..(seq.len() + 1).saturating_sub(k) {
                 //  Irradicate kmers containing 'N'
-                if !seq[i..i + k].contains(&n) {
-                    //  Make output kmers the lexicographically smaller of
-                    //  (kmer, reverse-complement)
+                if !seq[i..i + k].contains(&N) {
+                    //  Canonicalize by lexicographically smaller of kmer/reverse-complement
                     *fasta_hash
-                        .entry(min(seq[i..i + k].to_vec(), revcomp(&seq[i..i + k])))
+                        .entry(Box::from(min(&seq[i..i + k], &revcomp(&seq[i..i + k]))))
                         .or_insert(0) += 1;
                 } else {
                 }
@@ -73,14 +72,14 @@ pub fn run(config: Config) -> Result<(), Box<dyn Error>> {
 
     let mut buf = BufWriter::new(handle);
 
-    fasta_hash.into_iter().for_each(|(k, f)| {
+    fasta_hash.into_iter().for_each(|(kmer, f)| {
         //  Write:
         //  >frequency across fasta file for both kmer and its reverse complement
-        //  k-mer (lexicographically smaller of k-mer, reverse complement pair)
-        writeln!(buf, ">{}\n{}", f, str::from_utf8(&k).unwrap()).expect("Unable to write data");
+        //  canonical k-mer
+        writeln!(buf, ">{}\n{}", f, str::from_utf8(&kmer).unwrap()).expect("Unable to write data");
     });
-    
+
     buf.flush().unwrap();
-    
+
     Ok(())
 }