Mutect2 (normal BAM = FFPE-degraded, fragmented Low-DP)
I have run Mutect2 (matched) in 30 cases and 10 controls (brain DNA 60X is tumour sample and ffpe liver 30X is germline DNA). However some of my FFPE liver samples have bad coverage resulting in depths of around 10-15 and show fragmented reads.
I believe that Mutect2 will only remove variants at the germline comparison stage if the Alt shows evidence of presence in normal.. is it correct? So for example if a potential somatic SNV is identified in somatic sample and its corresponding loci in the normal sample lacks coverage (e.g. 0 depth) will the variant still be PASSed by Mutect2's first calling (before any filtermutectcalls or my own filtering) since there is no evidence of germline mutation present?
I want to check that a comparatively low callset of Somatic SNVs for a sample is not caused by low dp normal sample. Many Thanks :)
-
Hi Ben Thompson
As for your question, not finding any evidence in the normal mostly results in a PASS for something found in the somatic samples. However if your germline resource includes a site that was not covered in the normal and your somatic samples show that variant, it may get filtered out as Germline depending on the allele frequency in the germline resource file. So you may need to pay attention to germline resource allele frequencies if you get a germline filter for a known somatic variation.
I hope this helps.
Regards.
-
Hi Gökalp,
Thanks for your response it is very helpful. I assume it would be a rare occurrence but possible that mutect2 could mistakenly call somatic variant as germline if it is present at high VAF in population (germline resource) and missing site in normal sample.
for my data i am mostly interested in rate of somatic SNVs in disease group vs control rather than looking at known mutations so I believe the significant difference in somatic SNVs i have found is likely not due to ffpe degraded since coverage is mostly above 0 and likelihood of mutect2 calling somatic sample as germline seems quite small. Thanks very much for you help ! -
Hi again.
High population AF variants can be classified as potential germline events based on the math explained in the documentation below. We highly recommend you to check that before deciding to use the germline resource for your case.
https://github.com/broadinstitute/gatk/blob/master/docs/mutect/mutect.pdf
Regards.
Please sign in to leave a comment.
3 comments