Merge pull request #40 from bacpop/fastq_hotfix

Fastq input hotfix

Author: johnlees

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "sketchlib"
-version = "0.1.2"
+version = "0.1.3"
 authors = [
     "John Lees <[email protected]>",
     "Nicholas Croucher <[email protected]>",

src/distances.rs

@@ -137,7 +137,7 @@ impl<'a> Distances<'a> for DistanceMatrix<'a> {
     }
 }
 
-impl<'a> fmt::Display for DistanceMatrix<'a> {
+impl fmt::Display for DistanceMatrix<'_> {
     fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
         let mut dist_idx = 0;
         if let Some(queries) = &self.query_names {
@@ -267,7 +267,7 @@ impl<'a> Distances<'a> for SparseDistanceMatrix<'a> {
     }
 }
 
-impl<'a> fmt::Display for SparseDistanceMatrix<'a> {
+impl fmt::Display for SparseDistanceMatrix<'_> {
     fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
         let mut ref_name_iter = self.ref_names.iter();
         let mut ref_name = ref_name_iter.next().unwrap();

src/io.rs

@@ -89,7 +89,7 @@ pub fn get_input_list(
                     2 => ((fields[0].to_string()), fields[1].to_string(), None),
                     3 => (
                         (fields[0].to_string()),
-                        fields[0].to_string(),
+                        fields[1].to_string(),
                         Some(fields[2].to_string()),
                     ),
                     _ => {

src/multisketch.rs

@@ -15,7 +15,6 @@ use serde::{Deserialize, Serialize};
 use crate::hashing::HashType;
 use crate::sketch::{Sketch, BBITS};
 use crate::sketch_datafile::SketchArrayFile;
-use crate::utils::get_progress_bar;
 
 use std::collections::HashSet;
 #[derive(Serialize, Deserialize)]

src/structures.rs

@@ -1,7 +1,10 @@
 //! Support for .pdb files and the 3di alphabet
+#[cfg(feature = "3di")]
 use crate::io::InputFastx;
+#[cfg(feature = "3di")]
 use anyhow::Error;
 
+#[cfg(feature = "3di")]
 use indicatif::{ProgressIterator, ProgressStyle};
 #[cfg(feature = "3di")]
 use pyo3::prelude::*;

tests/common/mod.rs

@@ -115,27 +115,31 @@ impl TestSetup {
             writeln!(
                 rfile,
                 "{}",
-                &format!(
-                    "{}\t{}",
-                    file,
-                    self.file_string(file, TestDir::Input),
-                )
+                &format!("{}\t{}", file, self.file_string(file, TestDir::Input),)
             )
             .unwrap();
         }
         RFILE_NAME
     }
 
-    pub fn create_par_rfile(&self) -> &str {
+    /// Create an rfile wtih two fastqs in the tmp dir
+    pub fn create_fastq_rfile(&self, prefix: &str) -> &str {
         let mut rfile = LineWriter::new(
             File::create(format!("{}/{}", self.get_wd(), RFILE_NAME))
                 .expect("Could not write rfile"),
         );
-
-        let paths = read_dir(self.file_string("par_test", TestDir::Input)).unwrap();
-        for path in paths {
-            let path_str = path.unwrap().path().display().to_string();
-            writeln!(rfile, "{path_str}\t{path_str}").unwrap();
+        for file_idx in 1..=2 {
+            let sample_name = format!("{prefix}_{file_idx}");
+            writeln!(
+                rfile,
+                "{}",
+                &format!(
+                    "{sample_name}\t{}\t{}",
+                    self.file_string(&format!("{sample_name}_fwd.fastq.gz"), TestDir::Input),
+                    self.file_string(&format!("{sample_name}_rev.fastq.gz"), TestDir::Input),
+                )
+            )
+            .unwrap();
         }
         RFILE_NAME
     }

tests/delete.rs

@@ -16,7 +16,12 @@ mod tests {
     fn test_delete_sketches() {
         let sandbox = TestSetup::setup();
 
-        let rfile1 = sandbox.create_rfile(&vec!["14412_3#82.contigs_velvet.fa.gz", "14412_3#84.contigs_velvet.fa.gz", "R6.fa.gz", "TIGR4.fa.gz"]);
+        let rfile1 = sandbox.create_rfile(&vec![
+            "14412_3#82.contigs_velvet.fa.gz",
+            "14412_3#84.contigs_velvet.fa.gz",
+            "R6.fa.gz",
+            "TIGR4.fa.gz",
+        ]);
         Command::new(cargo_bin("sketchlib"))
             .current_dir(sandbox.get_wd())
             .arg("sketch")
@@ -27,7 +32,10 @@ mod tests {
             .assert()
             .success();
 
-        let rfile2 = sandbox.create_rfile(&vec!["14412_3#82.contigs_velvet.fa.gz", "14412_3#84.contigs_velvet.fa.gz"]);
+        let rfile2 = sandbox.create_rfile(&vec![
+            "14412_3#82.contigs_velvet.fa.gz",
+            "14412_3#84.contigs_velvet.fa.gz",
+        ]);
         Command::new(cargo_bin("sketchlib"))
             .current_dir(sandbox.get_wd())
             .arg("sketch")

tests/sketch.rs

@@ -0,0 +1,93 @@
+use predicates::prelude::*;
+use snapbox::cmd::{cargo_bin, Command};
+use std::path::Path;
+
+pub mod common;
+use crate::common::*;
+
+use sketchlib::multisketch::MultiSketch;
+
+#[cfg(test)]
+
+mod tests {
+    use super::*;
+
+    #[test]
+    fn sketch_fasta() {
+        let sandbox = TestSetup::setup();
+
+        // Create a fasta rfile in the tmp dir
+        Command::new(cargo_bin("sketchlib"))
+            .current_dir(sandbox.get_wd())
+            .arg("sketch")
+            .arg("-o")
+            .arg("assembly")
+            .args(["-v", "-k", "31"])
+            .arg(sandbox.file_string("14412_3#82.contigs_velvet.fa.gz", TestDir::Input))
+            .arg(sandbox.file_string("14412_3#84.contigs_velvet.fa.gz", TestDir::Input))
+            .assert()
+            .success();
+
+        assert_eq!(true, sandbox.file_exists("assembly.skm"));
+        assert_eq!(true, sandbox.file_exists("assembly.skd"));
+
+        Command::new(cargo_bin("sketchlib"))
+            .current_dir(sandbox.get_wd())
+            .arg("info")
+            .arg("assembly")
+            .assert()
+            .stdout_matches_path(
+                sandbox.file_string("assembly_sketch_info.stdout", TestDir::Correct),
+            );
+
+        Command::new(cargo_bin("sketchlib"))
+            .current_dir(sandbox.get_wd())
+            .arg("info")
+            .arg("--sample-info")
+            .arg("assembly")
+            .arg("-v")
+            .assert()
+            .stdout_matches_path(
+                sandbox.file_string("assembly_sketch_full_info.stdout", TestDir::Correct),
+            );
+    }
+
+    #[test]
+    fn sketch_fastq() {
+        let sandbox = TestSetup::setup();
+
+        // Create a fastq rfile in the tmp dir
+        let rfile_name = sandbox.create_fastq_rfile("test");
+        Command::new(cargo_bin("sketchlib"))
+            .current_dir(sandbox.get_wd())
+            .arg("sketch")
+            .arg("-f")
+            .arg(rfile_name)
+            .arg("-o")
+            .arg("reads")
+            .args(["--min-count", "2", "-v", "-k", "9", "--min-qual", "2"])
+            .assert()
+            .success();
+
+        assert_eq!(true, sandbox.file_exists("reads.skm"));
+        assert_eq!(true, sandbox.file_exists("reads.skd"));
+
+        Command::new(cargo_bin("sketchlib"))
+            .current_dir(sandbox.get_wd())
+            .arg("info")
+            .arg("reads")
+            .assert()
+            .stdout_matches_path(sandbox.file_string("read_sketch_info.stdout", TestDir::Correct));
+
+        Command::new(cargo_bin("sketchlib"))
+            .current_dir(sandbox.get_wd())
+            .arg("info")
+            .arg("--sample-info")
+            .arg("reads")
+            .arg("-v")
+            .assert()
+            .stdout_matches_path(
+                sandbox.file_string("read_sketch_full_info.stdout", TestDir::Correct),
+            );
+    }
+}

tests/test_files_in/test_1_fwd.fastq.gz

@@ -0,0 +1,4 @@
+Name	Sequence length	Base frequencies	Missing/ambig bases	From reads	Single strand	Densified
+[..]14412_3#82.contigs_velvet.fa.gz	1832266	[571702, 352572, 343740, 564252]	579	false	false	false
+[..]14412_3#84.contigs_velvet.fa.gz	1823561	[571676, 350424, 342380, 559081]	1663	false	false	false
+

tests/test_files_in/test_1_rev.fastq.gz

@@ -0,0 +1,5 @@
+sketch_version=[..]
+sequence_type=DNA
+sketch_size=1024
+n_samples=2
+kmers=[31]

tests/test_files_in/test_2_fwd.fastq.gz

@@ -0,0 +1,4 @@
+Name	Sequence length	Base frequencies	Missing/ambig bases	From reads	Single strand	Densified
+test_1	1	[603, 330, 334, 603]	0	true	false	true
+test_2	1	[617, 312, 314, 617]	0	true	false	true
+

tests/test_files_in/test_2_rev.fastq.gz

@@ -0,0 +1,5 @@
+sketch_version=[..]
+sequence_type=DNA
+sketch_size=1024
+n_samples=2
+kmers=[9]