chore: updates noodles to 0.34.0

Cargo.toml

@@ -19,7 +19,7 @@ git-testament = "0.2.1"
 indexmap = "1.9.1"
 indicatif = "0.16.2"
 itertools = "0.10.5"
-noodles = { version = "0.29.0", features = [
+noodles = { version = "0.34.0", features = [
     "async",
     "bam",
     "bgzf",
@@ -40,7 +40,7 @@ regex = "1.5.5"
 rust-lapper = "1.0.1"
 serde = { version = "1.0.137", features = ["derive"] }
 serde_json = { version = "1.0.81", features = ["preserve_order"] }
-tokio = { version = "1.18.0", features = ["fs", "rt-multi-thread"] }
+tokio = { version = "1.18.0", features = ["fs", "io-std", "rt-multi-thread"] }
 tracing = "0.1.34"
 tracing-subscriber = "0.3.11"
 

src/convert/bam.rs

@@ -1,6 +1,7 @@
 //! Conversions from a BAM file to other next-generation sequencing file formats.
 
 use std::io;
+use std::num::NonZeroUsize;
 use std::path::PathBuf;
 
 use anyhow::Context;
@@ -41,7 +42,7 @@ pub async fn to_sam_async(
     let mut record = Record::default();
 
     // (4) Write each record in the BAM file to the SAM file.
-    while reader.read_record(&mut record).await? != 0 {
+    while reader.read_record(&header.parsed, &mut record).await? != 0 {
         writer
             .write_alignment_record(&header.parsed, &record)
             .await?;
@@ -83,8 +84,8 @@ pub async fn to_cram_async(
         let name = name_as_string.parse()?;
         let length = record.sequence().len();
 
-        let reference_sequence = Map::<ReferenceSequence>::new(name, length)?;
-        reference_sequences.insert(name_as_string, reference_sequence);
+        let reference_sequence = Map::<ReferenceSequence>::new(NonZeroUsize::try_from(length)?);
+        reference_sequences.insert(name, reference_sequence);
     }
 
     let repository = fasta::Repository::new(records);
@@ -106,7 +107,7 @@ pub async fn to_cram_async(
 
     // (6) Write each record in the BAM file to the CRAM file.
     info!("Writing records to CRAM file.");
-    while reader.read_record(&mut record).await? != 0 {
+    while reader.read_record(&header.parsed, &mut record).await? != 0 {
         let cram_record = cram::Record::try_from_alignment_record(&header.parsed, &record)?;
         writer
             .write_record(&header.parsed, cram_record)

src/convert/sam.rs

@@ -1,6 +1,7 @@
 //! Conversions from a SAM file to other next-generation sequencing file formats.
 
 use std::io;
+use std::num::NonZeroUsize;
 use std::path::PathBuf;
 
 use anyhow::Context;
@@ -101,8 +102,8 @@ pub async fn to_cram_async(
         let name = name_as_string.parse()?;
         let length = record.sequence().len();
 
-        let reference_sequence = Map::<ReferenceSequence>::new(name, length)?;
-        reference_sequences.insert(name_as_string, reference_sequence);
+        let reference_sequence = Map::<ReferenceSequence>::new(NonZeroUsize::try_from(length)?);
+        reference_sequences.insert(name, reference_sequence);
     }
 
     let repository = fasta::Repository::new(records);

src/derive/command/instrument.rs

@@ -54,8 +54,9 @@ async fn app(src: PathBuf, first_n_reads: Option<usize>) -> anyhow::Result<()> {
     let mut instrument_names = HashSet::new();
     let mut flowcell_names = HashSet::new();
 
-    let ParsedBAMFile { mut reader, .. } =
-        crate::utils::formats::bam::open_and_parse(src, IndexCheck::Full)?;
+    let ParsedBAMFile {
+        mut reader, header, ..
+    } = crate::utils::formats::bam::open_and_parse(src, IndexCheck::Full)?;
 
     // (1) Collect instrument names and flowcell names from reads within the
     // file. Support for sampling only a portion of the reads is provided.
@@ -66,7 +67,7 @@ async fn app(src: PathBuf, first_n_reads: Option<usize>) -> anyhow::Result<()> {
         sample_max = s;
     }
 
-    for result in reader.records() {
+    for result in reader.records(&header.parsed) {
         let record = result?;
 
         if let Some(read_name) = record.read_name() {

src/index/bam.rs

@@ -72,7 +72,7 @@ pub fn index(src: PathBuf) -> anyhow::Result<()> {
     let mut counter = RecordCounter::new();
 
     loop {
-        match reader.read_record(&mut record) {
+        match reader.read_record(&header.parsed, &mut record) {
             Ok(0) => break,
             Ok(_) => {}
             Err(e) => bail!("failed to read record: {}", e),

src/qc.rs

@@ -7,6 +7,7 @@ use itertools::Itertools;
 use noodles::sam;
 use noodles::sam::header::record::value::map::ReferenceSequence;
 use noodles::sam::header::record::value::Map;
+use noodles::sam::record::ReferenceSequenceName;
 use noodles::sam::Header;
 use sam::alignment::Record;
 
@@ -200,14 +201,27 @@ pub trait SequenceBasedQualityControlFacet {
     fn supports_sequence_name(&self, name: &str) -> bool;
 
     /// Sets up a quality control facet for a given sequence.
-    fn setup(&mut self, sequence: &Map<ReferenceSequence>) -> anyhow::Result<()>;
+    fn setup(
+        &mut self,
+        name: &ReferenceSequenceName,
+        sequence: &Map<ReferenceSequence>,
+    ) -> anyhow::Result<()>;
 
     /// Processes a sequence for a quality control facet.
-    fn process(&mut self, seq: &Map<ReferenceSequence>, record: &Record) -> anyhow::Result<()>;
+    fn process(
+        &mut self,
+        name: &ReferenceSequenceName,
+        sequence: &Map<ReferenceSequence>,
+        record: &Record,
+    ) -> anyhow::Result<()>;
 
     /// Tears down any machinery that was built up for this sequence within the
     /// quality control facet.
-    fn teardown(&mut self, sequence: &Map<ReferenceSequence>) -> anyhow::Result<()>;
+    fn teardown(
+        &mut self,
+        name: &ReferenceSequenceName,
+        sequence: &Map<ReferenceSequence>,
+    ) -> anyhow::Result<()>;
 
     /// Adds the results of this quality control facet to the global
     /// [`results::Results`] object for writing to a file.

src/qc/command.rs

@@ -261,7 +261,7 @@ fn app(
         if !supported_sequences
             .iter()
             .map(|s| s.name())
-            .any(|x| x == sequence)
+            .any(|x| x == *sequence)
         {
             bail!(
                 "Sequence \"{}\" not found in specified reference genome. \
@@ -302,7 +302,7 @@ fn app(
         info!("Starting first pass for QC stats.");
         let mut counter = RecordCounter::new();
 
-        for result in reader.records() {
+        for result in reader.records(&header.parsed) {
             let record = result?;
 
             for facet in &mut record_facets {
@@ -361,22 +361,22 @@ fn app(
             debug!("    [*] Setting up sequence.");
             for facet in &mut sequence_facets {
                 if facet.supports_sequence_name(name) {
-                    facet.setup(seq)?;
+                    facet.setup(name, seq)?;
                 }
             }
 
             let query = reader.query(
-                header.parsed.reference_sequences(),
+                &header.parsed,
                 &index,
-                &Region::new(name, start..=end),
+                &Region::new(name.to_string(), start..=end),
             )?;
 
             debug!("    [*] Processing records from sequence.");
             for result in query {
                 let record = result?;
                 for facet in &mut sequence_facets {
                     if facet.supports_sequence_name(name) {
-                        facet.process(seq, &record)?;
+                        facet.process(name, seq, &record)?;
                     }
                 }
 
@@ -390,7 +390,7 @@ fn app(
             debug!("    [*] Tearing down sequence.");
             for facet in &mut sequence_facets {
                 if facet.supports_sequence_name(name) {
-                    facet.teardown(seq)?;
+                    facet.teardown(name, seq)?;
                 }
             }
         }