feat: anonymize bam - adapt mandatory and aux fields (#219)

* adapt mandatory fields

* make release-please happy

* remove variable

* move to distinct function

src/bam/anonymize_reads.rs

@@ -98,22 +98,17 @@ fn init_altered_bases(
 
 fn build_record(record: &bam::Record, artificial_seq: &[u8], offset: i64) -> Result<bam::Record> {
     let mut artificial_record = bam::record::Record::new();
-    if let Ok(mate_cigar) = record.aux(b"MC") {
-        artificial_record.push_aux(b"MC", mate_cigar)?;
-    }
     artificial_record.set(
         record.qname(),
         Some(&record.cigar()),
         artificial_seq,
         record.qual(),
     );
-    artificial_record.set_pos(record.pos() - offset);
-    artificial_record.set_tid(0);
-    artificial_record.set_mtid(0);
-    artificial_record.set_mpos(record.mpos() - offset);
-    artificial_record.set_flags(record.flags());
-    artificial_record.set_insert_size(record.insert_size());
-    artificial_record.set_mapq(record.mapq());
+    set_mandatory_fields(&mut artificial_record, record, offset)?;
+    for aux_result in record.aux_iter() {
+        let (tag, aux_field) = aux_result?;
+        artificial_record.push_aux(tag, aux_field)?;
+    }
     Ok(artificial_record)
 }
 
@@ -169,6 +164,28 @@ fn build_sequence(
     Ok(artificial_seq)
 }
 
+fn set_mandatory_fields(
+    target_rec: &mut bam::Record,
+    source_rec: &bam::Record,
+    offset: i64,
+) -> Result<()> {
+    target_rec.set_pos(source_rec.pos() - offset);
+    target_rec.set_tid(0);
+    let (mtid, mpos) = if source_rec.mtid() == -1 {
+        (-1, 0)
+    } else if source_rec.mtid() == source_rec.tid() {
+        (0, source_rec.mpos() - offset)
+    } else {
+        (1, source_rec.mpos())
+    };
+    target_rec.set_mtid(mtid);
+    target_rec.set_mpos(mpos);
+    target_rec.set_flags(source_rec.flags());
+    target_rec.set_insert_size(source_rec.insert_size());
+    target_rec.set_mapq(source_rec.mapq());
+    Ok(())
+}
+
 fn add_random_bases(
     length: u64,
     seq: &mut Vec<u8>,

src/bcf/match_variants.rs

@@ -119,7 +119,6 @@ pub fn match_variants<P: AsRef<Path>>(matchbcf: P, max_dist: u32, max_len_diff:
 #[derive(Debug)]
 pub struct Variant {
     id: u32,
-    rid: u32,
     pos: u64,
     alleles: Vec<VariantType>,
 }
@@ -206,7 +205,6 @@ impl Variant {
         };
         let var = Variant {
             id: *id,
-            rid: rec.rid().unwrap(),
             pos: pos as u64,
             alleles: _alleles,
         };

src/bcf/report/oncoprint.rs

@@ -987,8 +987,6 @@ struct Record {
     sample: String,
     gene: String,
     #[new(default)]
-    dna_alteration: Vec<String>,
-    #[new(default)]
     variants: Vec<String>,
 }
 

src/fastq/collapse_reads_to_fragments/pipeline.rs

@@ -63,7 +63,6 @@ fn isize_pmf(value: f64, mean: f64, sd: f64) -> LogProb {
 #[derive(Debug)]
 struct FastqStorage {
     db: DB,
-    storage_dir: std::path::PathBuf,
 }
 
 impl FastqStorage {
@@ -74,8 +73,7 @@ impl FastqStorage {
         // in turn deleting the tempdir
         let storage_dir = tempdir()?.path().join("db");
         Ok(FastqStorage {
-            db: DB::open_default(storage_dir.clone())?,
-            storage_dir,
+            db: DB::open_default(storage_dir)?,
         })
     }
 
@@ -351,11 +349,11 @@ impl<'a, R: io::Read + io::BufRead, W: io::Write> CallConsensusReads<'a, R, W>
     }
 
     fn fq1_reader(&mut self) -> &mut fastq::Reader<R> {
-        &mut self.fq1_reader
+        self.fq1_reader
     }
 
     fn fq2_reader(&mut self) -> &mut fastq::Reader<R> {
-        &mut self.fq2_reader
+        self.fq2_reader
     }
 
     fn umi_len(&self) -> usize {
@@ -506,11 +504,11 @@ impl<'a, R: io::Read + io::BufRead, W: io::Write> CallConsensusReads<'a, R, W>
     }
 
     fn fq1_reader(&mut self) -> &mut fastq::Reader<R> {
-        &mut self.fq1_reader
+        self.fq1_reader
     }
 
     fn fq2_reader(&mut self) -> &mut fastq::Reader<R> {
-        &mut self.fq2_reader
+        self.fq2_reader
     }
 
     fn umi_len(&self) -> usize {