update dependencies and clean up code in genotype and parse_bam modules

Cargo.toml

@@ -8,27 +8,27 @@ edition = "2021"
 
 [dependencies]
 clap = { version = "3.2.6", features = ["derive"] }
-env_logger = "0.11.3"
-flate2 = "1.0.24"
-log = "0.4.17"
+env_logger = "0.11.6"
+flate2 = "1.0.35"
+log = "0.4.25"
 rayon = "1.5.3"
-rust-htslib = { version = "0.47.0", features = ["s3", "curl"] }
+rust-htslib = { version = "0.49.0", features = ["s3", "curl"] }
 bio = { git = "https://github.com/lorewar2/rust-bio.git", branch = "reverse_edge_fix" }
 human-sort = "0.2.2"
 minimap2 = { version = "0.1.20", features = ["htslib"] }
 kodama = "0.3.0"
-regex = "1.8.1"
-url = "2.3.1"
+regex = "1.11.1"
+url = "2.5.4"
 distance = "0.4.0"
 levenshtein = "1.0.5"
 rand = "0.8.5"
-libz-sys = "1.1.12"
+libz-sys = "1.1.21"
 libc = "0.2.147"
-petgraph = "0.6.4"
+petgraph = "0.7.1"
 hts-sys = "2.1.1"
 reqwest = { version = "0.12.5", features = ["blocking", "json"] }
-indicatif = { version = "0.17.1", features = ["rayon"] }
-chrono = "0.4.38"
+indicatif = { version = "0.17.9", features = ["rayon"] }
+chrono = "0.4.39"
 
 [dev-dependencies]
 ctor = "*"

src/genotype.rs

@@ -54,13 +54,6 @@ fn genotype_repeat(
     };
 
     let repeat_compressed_reference = repeat.make_repeat_compressed_sequence(&args.fasta, flanking);
-    // if args.debug {
-    //     // write the repeat compressed reference to a file
-    //     use std::fs;
-    //     let header = format!(">{chrom}:{start}-{end}\n", chrom = repeat.chrom, start = repeat.start, end = repeat.end);
-    //     let fas = String::from_utf8(repeat_compressed_reference.clone()).expect("Unable to convert repeat compressed reference to string");
-    //     fs::write("repeat_compressed.fa", header + &fas).expect("Unable to write repeat compressed reference to file");
-    // }
 
     // alignments can be extracted in an unphased manner, if the chromosome is --haploid or the --unphased is set
     // this means that --haploid overrides the phases which could be present in the bam file
@@ -113,7 +106,7 @@ fn genotype_repeat(
             .collect::<Vec<&str>>()
             .contains(&repeat.chrom.as_str())
     {
-        // if the chromosome is haploid, all reads are put in phase 0
+        // if the chromosome is haploid, all reads were put in phase 0
         let seq = reads.seqs.get(&0).unwrap();
         debug!("{repeat}: Haploid: Aligning {} reads", seq.len());
         let insertions = find_insertions(seq, &aligner, args.minlen, flanking, repeat);
@@ -254,7 +247,7 @@ fn genotype_repeat(
 // may adapt the function below to allow for multiple alignment methods later
 fn find_insertions(
     seq: &Vec<Vec<u8>>,
-    aligner: &Aligner,
+    aligner: &Aligner<Built>,
     minlen: usize,
     flanking: u32,
     repeat: &crate::repeats::RepeatInterval,
@@ -263,7 +256,7 @@ fn find_insertions(
 
     // align the reads to the new repeat-compressed reference
     for s in seq {
-        let mapping = aligner.map(s.as_slice(), true, false, None, None).unwrap_or_else(|err| panic!("Unable to align read with seq {s:?} to repeat-compressed reference for {repeat}\n{err}", s=s.to_ascii_uppercase()));
+        let mapping = aligner.map(s.as_slice(), true, false, None, None, None).unwrap_or_else(|err| panic!("Unable to align read with seq {s:?} to repeat-compressed reference for {repeat}\n{err}", s=s.to_ascii_uppercase()));
         for read in mapping {
             if let Some(s) = parse_cs(read, minlen, flanking, repeat) {
                 // slice out inserted sequences from the CS tag
@@ -382,7 +375,7 @@ mod tests {
             .with_cigar()
             .with_seq(&repeat_compressed_reference)
             .expect("Unable to build index");
-        let mapping = aligner.map(read.as_slice(), true, false, None, None).unwrap_or_else(|_| panic!("Unable to align read with seq {read:?} to repeat-compressed reference for {repeat}", read=read.to_ascii_uppercase()));
+        let mapping = aligner.map(read.as_slice(), true, false, None, None, None).unwrap_or_else(|_| panic!("Unable to align read with seq {read:?} to repeat-compressed reference for {repeat}", read=read.to_ascii_uppercase()));
 
         let _insertion = parse_cs(
             mapping

src/parse_bam.rs

@@ -96,7 +96,7 @@ pub fn get_overlapping_reads(
         for (phase, seqs_phase) in seqs.iter() {
             let n_reads = seqs_phase.len();
             let n_reads_to_select = if n_reads > max_reads_per_phase[&phase] {
-                max_reads_per_phase[&phase]
+                max_reads_per_phase[phase]
             } else {
                 n_reads
             };

src/repeats.rs

@@ -179,17 +179,13 @@ impl RepeatInterval {
                 (self.end + flanking - 2) as usize,
             )
             .expect("Failed to extract fas_right sequence from fasta for {chrom}:{start}-{end}");
-
-        let newref = [fas_left, fas_right].concat();
-        unsafe { libc::free(fas_left.as_ptr() as *mut std::ffi::c_void) }; // Free up memory (https://github.com/rust-bio/rust-htslib/issues/401#issuecomment-1704290171)
-        unsafe { libc::free(fas_right.as_ptr() as *mut std::ffi::c_void) }; // Free up memory
-        newref
+        [fas_left, fas_right].concat()
     }
 
     pub fn reference_repeat_sequence(&self, fasta: &String) -> Option<String> {
         let fas = faidx::Reader::from_path(fasta).expect("Failed to read fasta");
         let repeat_ref_sequence = std::str::from_utf8(
-            fas.fetch_seq(&self.chrom, self.start as usize - 1, self.end as usize)
+            &fas.fetch_seq(&self.chrom, self.start as usize - 1, self.end as usize)
                 .expect("Failed to extract repeat sequence from fasta for {chrom}:{start}-{end}"),
         )
         .expect("Failed to convert repeat sequence to string for {chrom}:{start}-{end}")