Filtering somatic calls from tumor-only variant calling
[GATK version used - v220.127.116.11]
I have germline and somatic variants called using HaplotypeCaller and Mutect2. As you can see from the title of the posting, it was done using tumor samples without matched normals.
I was wondering what would be "best" (since matched normal would be actually best) approach to filter out false positive calls from Mutect2. My plan originally was to first filter out any variants in germline and somatic calls that overlap, if there are any.
Would it be better to look into other tools that specializes in looking for somatic variants when only tumor samples are available?
Run Mutect2 and FilterMutectCalls with the best practices resources for germline resource and panel of normals. The Mutect2 pipeline is self-contained.
How confident are calls retained using the option '--genotype-germline-sites' for the germline variants (since it is said to be experimental)? I've browsed through some older posts, but haven't been able to come to a good conclusion.
By default, Mutect2 skips obvious germline variants and only sends cases to mild doubt to FilterMutectCalls. --genotype-germline-sites disables that optimization and tells Mutect2 to go through the expense of assembly and realignment for every site. This means that your output VCF will have records for all germline variants. It does not means that Mutect2 tries to do full germline genotyping like HaplotypeCaller. It is important to keep in mind that Mutect2 is only interested in germline sites insofar as they are not somatic variants. If a site is flagged as "germline", it means FilterMutectCalls thinks it's a bad idea to trust it in a somatic callset, but it should not necessarily be considered a confident germline call.
Please sign in to leave a comment.