Filter MuTect before running MutSig?

Question

[simonjean434]

Hello all,

Should we filter (clean) the somatic variants called by MuTect before running MutSig?

If yes, which filtering criteria you apply?

Thanks
J

[poisonAlien]

What do you mean by filtering somatic variants before mutsig? Are asking about removing synonymous and intronic variants ? Or removing false positive variants ?

[simonjean434]

Thanks for your reply. Actually, both! also should we remove alternative variants with low coverage?

[poisonAlien]

It's always good to remove low coverage variants since it improves the estimated vaf. Also given MuTect detects lot of subclonal mutations with low vaf's, it is desirable to choose those with reliable depth. And for second, MutSig only considers variants which affect the structure/function of the protein (non-synonymous), to pick out drivers. So it should not affect the analysis much. In fact if you are working on liquid tumors (such as AML) where mutation rate is low (8-10 per sample), it is good to include those tier 2 and tier 3 mutations, just to improve the statistics behind MutSig.

You can check this for effect of coverage on vafs and variant consideration.

[simonjean434]

Thank you for your help!