Illumina / manta

Structural variant and indel caller for mapped sequencing data
GNU General Public License v3.0
404 stars 154 forks source link

Filter bad-cigar reads in bam streamer #288

Open fbattke opened 2 years ago

fbattke commented 2 years ago

These reads are ok according to SAM/BAM spec, but Manta can't work with them.

This fixes issue https://github.com/Illumina/manta/issues/137 "Manta crashes on input files realigned with ABRA #137" as well as issue https://github.com/Illumina/manta/issues/184 "Unclear how to deal with CIGAR strings with InDels at the end of reads"

The fix is to remove offending alignments in the bam streamer instead of throwing an exception much later in the code. This has the same effect as pre-filtering the alignments using awk (as described in issue 184), but it is much faster.

fbattke commented 2 years ago

We have tested this change on 100 cases and all output was identical to the original workaround using awk to filter alignments before calling manta.