`FilterMutectCalls` for normal only calls
Answered
Why do I see that the guide for `CreateSomaticPanelOfNormals` is `FilterMutectCalls` not used after calling germline `Mutect2` variants? Is this just a suggestion, and how strong is that suggestion? If a set of variants in my panel of normals have a clustered event, I don't want that to filter out any non-clustered events in the same locus when I run my tumor.
-
Hi AMN,
The panel of normals is meant to be a source listing technical artifacts from the normals (source). Thus, the VCF should not be filtered after running Mutect2. We have done rigorous testing on the method to create the panel of normals and this is the method that gets the best reliable results with Mutect2.
Best,
Genevieve
-
That's quite helpful, thanks! I am still concerned about cases like the one I mentioned, but it is good to know that overall this gave you the best results.
-
Glad it was helpful! I will keep your question regarding the clustered events in our backlog to ask the developers. There are a few events in the next few weeks so I probably won't get an answer soon.
-
Hi AMN,
I followed up with our developer team regarding your question. Yes, an event in your tumor that is within the clustered event in your PON would get filtered out. Unfortunately, this is probably still the best way to handle somatic variant calling since there are so many artifacts in tumor samples and not many true somatic variants.
You can manually remove these sites if you think they are removing sites of true variation or adjust the sensitivity parameters. You will probably then get an issue of false positives appearing in your output.
Best,
Genevieve
Please sign in to leave a comment.
4 comments