M2 is not calling a validated variant in a Coriell reference sample (NA14634): a 4bp deletion in BRCA1 exon 10 with an expected 50% AF, which can be easily seen in the bam with IGV. The bamout does not show the reads with the 4bp deletion so M2 is either filtering out those reads with the deletion or realigning them in a way that the 4bp deletion is lost.
Comparing the original bam with the bamout, there is a clear difference in coverage (nearly half the cov. in the bamout) even when I include the `--disable-tool-default-read-filters true` argument. I already posted on the old forum that some disable read arguments in M2 seem to be not working. I am also removing the coverage cap with the argument `--max-reads-per-alignment-start 0` so that is not the source of this difference in coverage.
This is data from amplicon tech, as previously discussed in my other posts.
I am adding snapshots showing the site in the bam and in the bamout, respectively
In case this info could be useful: this variant is correcly called by HaplotypeCaller yet only if softclips (from trimmed primers) are removed from the bam file. I am adding an independent post on this other issue soon.
I am using GATK 22.214.171.124 (latest release)
Thank you in advance
Please sign in to leave a comment.