Is it possible to extract somatic variants from paired VCF files?

Question

[csardas]

Hi all,

I have 2 paired VCF files (normal and tumor sample). Is it possible to extract somatic variants from VCF directly? or I have to go back to bam files and run Mutect or VarScan?

[WouterDeCoster]

So essentially you would want to find variants which are present in the tumor.vcf and not in the normal.vcf?

[Sandeep]

How did you finally end up doing it?

Answer 1

[biomaster]

Yes, it is possible. Two ways:
1, the simple way, check the frequency for the mutations exist in both, if normal AF is close to 0.5 or 1.0, and tumor AF is more or less, then it is germline.
2, better way, but more complicated. Use the allele count and depth of the mutations exist in both, and then do a t-test, calculate the p-value.