Call germline mutations from Mutect2
Is it possible to call germline mutations with Mutect2 (in addition to somatic variants) ?
Because I read on the net, that the algorithm of Mutect2 will eliminate germline variants and I have to use the Haplotype caller instead.
If it's true, I have 2 questions:
1) How do the Mutect2 algorithm distinguish between Germline and Somatic variant ? (without using an external database like GnomAD). Is it based only on the AF ?
2) What if I have a sample with a large proportion of tumoral cells (so my somatic variant have a high AF, like 80%)
Thanks in advance
Its recommended that you use Haplotypecaller for germline variants and Mutect2 for somatic variants.
Aside from population resources like gnomad "Mutect2 uses the matched normal to additionally exclude rare germline variation not captured by the germline resource and individual-specific artifacts. [Also,] A panel of normals (PoN) has a vital role that fills a gap between the matched normal and the population resource. Mutect2 uses the PoN to catch additional sites of noise in sequencing data, like mapping artifacts or other somewhat random but systematic artifacts of sequencing and data processing."
So If I 'm not using, neither Matched Normal nor PoN nor Population ressources, there is no reason for Mutect2 to miss germline variants. Do you agree ?
Chadi Saad In theory, yes, but keep in mind that Mutect2 is interested in flagging germine mutations, not calling them. This means that it doesn't try to filter germline variants for errors. Also, you will need the `--genotype-germline-sites` flag to avoid silently skipping obvious germline variants. As Beri said, HaplotypeCaller is a better tool for germline variation.
Q1: Please take a look at sections IIIC and IIIE of the supplementary material of this paper: https://www.biorxiv.org/content/10.1101/861054v1.
Q2: Mutect2 has no problem calling high-AF somatic variants, and the germline filter will still work in such cases.
Please sign in to leave a comment.