Why so many FP indels reported when running HaplotypeCaller on Indels for SOLiD data

ledwardsledwards LondonPosts: 14Member

We ran a recent version of Haplotyper Caller on our SOLiD targeted resequencing data and got a ridiculous number of indels. We took a closer look at some and there was absolutely no evidence for an indel at a called position, and wondered whether the internal realignment was doing something weird? Is this a known problem for SOLiD data? Our Illumina data works much better. It makes us now wary of using GATK for SOLiD data...is it just a filtering thing?

Best Answer

Answers

  • Geraldine_VdAuweraGeraldine_VdAuwera Posts: 6,462Administrator, GATK Developer admin

    We don't work with SOLiD data at all so it's possible there is a specific issue here that we're not aware of. One thing you can do to figure this out is use the bamOut argument to generate the realigned BAM for a given region, which will tell you what the HC thinks is going on in there. If it makes no sense then you know that the HC is doing the wrong thing, in which case you'll probably need to stick with UG for that data. Please let us know what you find so we know whether we need to add a caveat to the documentation and help other users avoid this issue. Sorry for the inconvenience!

    Geraldine Van der Auwera, PhD

  • ledwardsledwards LondonPosts: 14Member

    Very helpful thank you. Would it be wrong to call SNPs using HC and Indels using UG - or does the fact that it does weird things with HC for Indels mean that it could also be unreliable for SNPs?

Sign In or Register to comment.