I recently ran MutSigCV on 6 tumor-germline pairs for esophageal adenocarcinoma (EAC), but MutsigCV could not find any significantly mutated genes. For example, TP53 which has previously been implicated in EAC wasn't significantly mutated according to Mutsig, even though TP53 was mutated across all the tumors.
What could probably be causing this? Small sample size? The way the data was filtered before running Mutsig?
I filtered the data as follows :
- Tumor sample minus corresponding Normal sample
- esp, 1000genomes set at 0.01
- var % > 0.2
- exonic, exonic;splicing, splicing
- minimum of 10 reads covering a site in the tumor and normal sample
Please, I really could use someone's help right now.
•
written
by
murat.arslan13 •
3 answers
I suspect sample size is the isssue: I think MutSig and MuSiC are really meant to be run on cohorts with 100s of patients.
Also, just to be clear, you should really be using a somatic variant caller (such as varscan somatic or MuTect). I mention this because you say "tumor sample minus corresponding normal sample," and you don't want to just do a venn diagram to define mutants that are only present in the tumor sample. For example, the coverage may vary at those sites and/or you may have variant marginally do and do not meet the criteria for significance (which would look different in the venn diagram but are actually have similar signals).
•
written
by
Charles Warden
Thank you for your prompt reply. For this reason I've decided not to use mutsig. I'll be using another program to uncover driver genes, probably Oncodrive.
Hello,
in the same context, I used Oncodrive clust and FM to identify potential driver genes. However OncodriveClust did not report any significant genes..
My samples are 16 and total number of unique mutation > 60k (47% nonSYN, 53% SYN). Could that be due to the relatively more SYN mutations, since OncodriveClust use them as a background?
Any suggestion how can I polish such outcome?
Thank you in advance
Log in to answer this question.
