We've moved!
This site is now read-only. You can find our new documentation site and support forum for posting questions here.
Be sure to read our welcome blog!

Outputting bamout to read post-realignment coverage including non-variants


I want to output a bamout to see the post-realignment coverage (a la -A DepthPerAlleleBySample) for a position, regardless of whether that site is a variant site. I'm sure I'm need to turn on -disableOptimizations and --emitDroppedReads, but should I also turn on --forceActive and --dontTrimActiveRegions? In addition, would these flags affect the results of the variant calling itself?

Thanks for the help!


Best Answers


  • valentinvalentin Cambridge, MAMember, Dev ✭✭

    I think that --dontTrimActiveRegions may well change the results like the PL/QUAL values somewhat however only in rare cases it will change GT call or allele list would be rare though.

  • johnmajohnma Member

    But I thought AD counts unfiltered reads as well. Aren't informative, but filtered, reads not output to standard bamout?

    In addition, I have looked at the source code, and noticed the possible reasons for being tagged as dropped include Trimmed, Downsampled, Failed realignment, Filtered, and Not informative. I suppose only Failed realignment and Not informative reads won't be counted into allele, right?

  • valentinvalentin Cambridge, MAMember, Dev ✭✭

    For a read to count towards AD at a site, it must overlap the variant after realignment and it must have exhibit some difference in likelihoods between the top two alleles (the so called "informative" reads).

    I think the user may ask for with some option, which I don't remember right now, to use filtered out (and perhaps trimmed?) reads for calculating annotations (but never for PL/GQ or QUAL), but the way they are re-added they won't be considered informative anyway.

    Down sampled reads won't ever be considered for AD or another annotation and I reckon that they will be displayed with the original alignment.

  • johnmajohnma Member

    I take that to mean -disableOptimizations --forceActive --dontTrimActiveRegions should emit all realigned reads that are used to count AD, including in non-variant regions. Is that right?

Sign In or Register to comment.