added feature per clément to output count file for each read file

TravisWheelerLab · Jun 28, 2024 · f60eb67 · f60eb67
1 parent d46430a
commit f60eb67
Show file tree

Hide file tree

Showing 14 changed files with 73 additions and 525 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/Cargo.toml b/Cargo.toml
@@ -10,6 +10,7 @@ anyhow = "1.0.86"
 kseq = "0.5"
 clap = { version = "4.5.7", features = ["derive"] }
 rayon = "1.10.0"
+itertools = "0.13.0"
 
 [dev-dependencies]
 assert_cmd = "2"

diff --git a/mk-outs.sh b/mk-outs.sh
@@ -17,8 +17,10 @@ OUT_FA_100K="tests/outputs/out-100k-fasta.txt"
 OUT_FQ_50K="tests/outputs/out-50k-fastq.txt"
 OUT_FQ_100K="tests/outputs/out-100k-fastq.txt"
 
-$PRG -d $DNA_FA -r $RNA_FA_50K  -o $OUT_FA_50K
-$PRG -d $DNA_FA -r $RNA_FA_100K -o $OUT_FA_100K
+OUT_DIR="tests/outputs"
 
-$PRG -d $DNA_FA -r $RNA_FQ_50K  -o $OUT_FQ_50K
-$PRG -d $DNA_FA -r $RNA_FQ_100K -o $OUT_FQ_100K
+$PRG -j $DNA_FA -r $RNA_FA_50K  -o $OUT_DIR
+$PRG -j $DNA_FA -r $RNA_FA_100K -o $OUT_DIR
+
+$PRG -j $DNA_FA -r $RNA_FQ_50K  -o $OUT_DIR
+$PRG -j $DNA_FA -r $RNA_FQ_100K -o $OUT_DIR
diff --git a/src/main.rs b/src/main.rs
@@ -109,22 +109,30 @@ fn run(args: Args) -> Result<()> {
         .try_for_each(|reads_file| -> Result<()> {
             let basename = Path::new(&reads_file)
                 .file_name()
-                .ok_or(anyhow!("basename"))?;
+                .ok_or(anyhow!("basename"))?
+                .to_os_string();
 
-            let mut basename = basename.to_os_string();
-            basename.push(".txt");
-            let out_path = &outdir.join(basename);
-            let mut out_file = File::create(out_path)?;
+            let mut out_data_file = basename.clone();
+            out_data_file.push(".txt");
+            let out_data_path = &outdir.join(out_data_file);
+            let mut out_data = File::create(out_data_path)?;
+
+            let mut out_count_file = basename.clone();
+            out_count_file.push(".count");
+            let out_count_path = &outdir.join(out_count_file);
+            let mut out_count = File::create(out_count_path)?;
 
             // Search through each of the RNA sequences, reusing
             // the sequence and search results instances.
             let timer = Instant::now();
             let mut reads: kseq::Paths = get_reader(&reads_file)?;
-            writeln!(out_file, "File: {}", &reads_file)?;
+            writeln!(out_data, "File: {}", &reads_file)?;
 
             let mut search: Search = Search::new(&junctions)?;
+            let mut read_count = 0;
             while let Some(rec) = reads.iter_record()? {
                 search.search(rec.seq());
+                read_count += 1;
             }
 
             if args.verbose {
@@ -137,10 +145,12 @@ fn run(args: Args) -> Result<()> {
             for (i, count) in search.junctions.hits.into_iter().enumerate() {
                 if count > 0 {
                     if let Some(name) = map.get(&search.junctions.key[i]) {
-                        writeln!(out_file, "{name}\t{count}")?;
+                        writeln!(out_data, "{name}\t{count}")?;
                     }
                 }
             }
+
+            writeln!(out_count, "{read_count}")?;
             Ok(())
         })?;
 

diff --git a/tests/cli.rs b/tests/cli.rs
@@ -1,22 +1,30 @@
 use anyhow::{anyhow, Result};
 use assert_cmd::Command;
+use itertools::izip;
 use predicates::prelude::*;
 use pretty_assertions::assert_eq;
 use rand::{distributions::Alphanumeric, Rng};
-use std::{fs, iter::zip, path::Path};
+use std::{fs, path::Path};
 use tempfile::TempDir;
 
 const PRG: &str = "tallyman";
 const DNA_FA: &str = "tests/inputs/dna.fasta";
 const DNA_FQ: &str = "tests/inputs/dna.fastq";
+
 const RNA_FA_50K: &str = "tests/inputs/rna-50k.fasta";
 const RNA_FQ_50K: &str = "tests/inputs/rna-50k.fastq";
 const RNA_FA_100K: &str = "tests/inputs/rna-100k.fasta";
 const RNA_FQ_100K: &str = "tests/inputs/rna-100k.fastq";
-const OUT_FA_50K: &str = "tests/outputs/out-50k-fasta.txt";
-const OUT_FA_100K: &str = "tests/outputs/out-100k-fasta.txt";
-const OUT_FQ_50K: &str = "tests/outputs/out-50k-fastq.txt";
-const OUT_FQ_100K: &str = "tests/outputs/out-100k-fastq.txt";
+
+const OUT_FA_50K: &str = "tests/outputs/rna-50k.fasta.txt";
+const OUT_FA_100K: &str = "tests/outputs/rna-100k.fasta.txt";
+const OUT_FA_50K_COUNT: &str = "tests/outputs/rna-50k.fasta.count";
+const OUT_FA_100K_COUNT: &str = "tests/outputs/rna-100k.fasta.count";
+
+const OUT_FQ_50K: &str = "tests/outputs/rna-50k.fastq.txt";
+const OUT_FQ_100K: &str = "tests/outputs/rna-100k.fastq.txt";
+const OUT_FQ_50K_COUNT: &str = "tests/outputs/rna-50k.fastq.count";
+const OUT_FQ_100K_COUNT: &str = "tests/outputs/rna-100k.fastq.count";
 
 // --------------------------------------------------
 fn gen_bad_file() -> String {
@@ -64,9 +72,11 @@ fn run(
     read_files: &[&str],
     junction_file: &str,
     expected_files: &[&str],
+    expected_counts: &[&str],
 ) -> Result<()> {
     // outdir will be removed when var leaves scope
     let outdir = TempDir::new()?;
+
     let mut args: Vec<String> = vec![
         "-j".to_string(),
         junction_file.to_string(),
@@ -81,19 +91,30 @@ fn run(
 
     Command::cargo_bin(PRG)?.args(&args).assert().success();
 
-    for (read_file, expected_file) in zip(read_files, expected_files) {
+    for (read_file, expected_file, expected_count) in
+        izip!(read_files, expected_files, expected_counts)
+    {
         // Output file is read basename + ".txt"
-        let mut read_base = Path::new(&read_file)
+        let read_base = Path::new(&read_file)
             .file_name()
             .ok_or(anyhow!("No basename"))?
             .to_os_string();
-        read_base.push(".txt");
-        let outpath = &outdir.path().join(&read_base);
+        let mut data_basename = read_base.clone();
+        data_basename.push(".txt");
+        let outpath = &outdir.path().join(&data_basename);
         assert!(outpath.exists());
 
         let expected = fs::read_to_string(expected_file)?;
         let actual = fs::read_to_string(outpath)?;
         assert_eq!(&actual, &expected);
+
+        let mut count_basename = read_base.clone();
+        count_basename.push(".count");
+        let outpath = &outdir.path().join(&count_basename);
+        assert!(outpath.exists());
+        let expected = fs::read_to_string(expected_count)?;
+        let actual = fs::read_to_string(outpath)?;
+        assert_eq!(&actual, &expected);
     }
 
     Ok(())
@@ -102,25 +123,25 @@ fn run(
 // --------------------------------------------------
 #[test]
 fn run_50k_fasta() -> Result<()> {
-    run(&[RNA_FA_50K], DNA_FA, &[OUT_FA_50K])
+    run(&[RNA_FA_50K], DNA_FA, &[OUT_FA_50K], &[OUT_FA_50K_COUNT])
 }
 
 // --------------------------------------------------
 #[test]
 fn run_50k_fastq() -> Result<()> {
-    run(&[RNA_FQ_50K], DNA_FQ, &[OUT_FQ_50K])
+    run(&[RNA_FQ_50K], DNA_FQ, &[OUT_FQ_50K], &[OUT_FQ_50K_COUNT])
 }
 
 // --------------------------------------------------
 #[test]
 fn run_100k_fasta() -> Result<()> {
-    run(&[RNA_FA_100K], DNA_FA, &[OUT_FA_100K])
+    run(&[RNA_FA_100K], DNA_FA, &[OUT_FA_100K], &[OUT_FA_100K_COUNT])
 }
 
 // --------------------------------------------------
 #[test]
 fn run_100k_fastq() -> Result<()> {
-    run(&[RNA_FQ_100K], DNA_FQ, &[OUT_FQ_100K])
+    run(&[RNA_FQ_100K], DNA_FQ, &[OUT_FQ_100K], &[OUT_FQ_100K_COUNT])
 }
 
 // --------------------------------------------------
@@ -130,5 +151,6 @@ fn run_50k_100k_fastq() -> Result<()> {
         &[RNA_FA_50K, RNA_FQ_100K],
         DNA_FQ,
         &[OUT_FA_50K, OUT_FQ_100K],
+        &[OUT_FA_50K_COUNT, OUT_FQ_100K_COUNT],
     )
 }