Indel Realignment in Tumor-Normal pairs

MilanDMilanD BelgradeMember

Hello dear GATK team,

I have a question regarding indel realignment when doing somatic variant calling (tumor normal pairs). First of all, I am aware that indel realignment is not required by mutect2 but I am still using it for other somatic variant callers. There are a couple of papers mentioning that indel realignment should be done jointly (with both BAMs as input of the RealignerTargetCreator amd IndelRealigner). I tested this using simulated datasets and it did give better results.

I am interested in the reasoning for this, and is this something that should happen in general? In addition, I have a concern regarding differences in coverage. Would this method cause issues if there is a significantly higher coverage in a tumor sample to that in a normal sample?

Thanks in advance!



  • SheilaSheila Broad InstituteMember, Broadie, Moderator admin

    Hi Milan,

    We recommend running Indel Realignment on both the tumor and normal sample together so that the tumor and normal have the same events, which helps in eliminating germline events.

    I don't think differences in coverage should affect indel realignment. The tools only perform realignment if there is evidence in the reads for an indel. Higher coverage may help to uncover an indel, but both samples will undergo the realignment because the tools ignore all read group information and simply process all reads together.


Sign In or Register to comment.