Test-drive the GATK tools and Best Practices pipelines on Terra

Check out this blog post to learn how you can get started with GATK and try out the pipelines in preconfigured workspaces (with a user-friendly interface!) without having to install anything.

PoN causing missed somatic variant call with Mutect2

amfgcpamfgcp Member
edited July 2 in Ask the GATK team
Hi, I've been running Mutect2 for somatic variant calling and I decided to created a panel of normals (PoN) using "CreateSomaticPanelOfNormals". I followed the instructions for GATK as that is the version I've been using.

When going through the differences in the somatic variants called by Mutect2 with and without the PoN I noticed a case I was not expecting: using the PoN, Mutect2 did not call a previously called somatic variant because it was located in the same position as one of the germline variants caught by the PoN (present in 4 of the ~20 samples used in its creation). I tried re-running Mutect2 with the parameter "--genotypePonSites true" hoping that maybe for this case the variant would not be filtered but as expected by the parameter description it was.

I'll leave a screenshot of IGV where on the bottom track is the missed somatic variant and on the track above that is the germline variant (present in 3 other samples that were also used to create the PoN).

I have checked the release notes for newer GATK versions but I found none that seemed to address this case. Could I be missing something? Is the mentioned behaviour expected? If it is, is there a plan to change it in a future release?



  • amfgcpamfgcp Member
    Re-reading my post I think I should be a bit clearer, the issue is that for the same position the somatic variant is a different allele than that of the germline variant but it ends up being filtered anyway
  • amfgcpamfgcp Member
    edited July 15
    Using "-germline-resource af-only-gnomad.vcf" in "CreateSomaticPanelOfNormals" solved the mentioned case, but in theory I think it is still possible for it occur so I would be grateful if there was any info on this :)
    Also, I am now using gatk, but I believe the only meaningful difference is that before the "-germline-resource" parameter was passed in calling Mutect2 in tumor-only mode for each of the normal samples (1st step) when creating the PoN.
Sign In or Register to comment.