We've moved!
This site is now read-only. You can find our new documentation site and support forum for posting questions here.
Be sure to read our welcome blog!

Comparison of pairs of tumor samples

Dear GATK, I would like to analyse data from pairs of samples. Each pair is a sample from a tumor and from a cell line derived from that tumor. We expect the tumor samples to be quite heterogenous, and the derived cell lines be less heterogenous. I want to find what differs between the tumor and the derived cell line. In general we think most of the variants present in the derived cell lines will also present in the original tumor (although possibly at a different frequency). Of primary interest would be variants present in tumor but absent ( or present at a reduced frequency ) in the cell lines.
I thought I could either:
1) Run mutect2 in matched pair mode, using the cell lines as the normal samples. I would have to set ploidy suitably high to account for low frequency variants. I think I could omit the PON as I don't think it is very important whether or not the the variant is likely to be germline, and anyway I don't have normal samples. It is possible that the "normal" sample in this scenario will contain large deletions not found in in the "tumour" samples, will this cause problems?
2) Run mutect2 in tumor only mode, then compare the output using another tool such as "GenotypeConcordance"
Do either of these options seem sensible? Also is it possible to get mutect2 to output all sites (not just variant sites).
Thank you and sorry for the long question



  • SheilaSheila Broad InstituteMember, Broadie ✭✭✭✭✭

    Hi Frances,

    I want to find what differs between the tumor and the derived cell line.

    In that case, it seems like option 1 may be the best. This will output sites where the "normal" does not have variants, but the tumor does.Are you looking for somatic mutations only? If so, you should include the PoN and germline resource which will help exclude possible germline variants.

Sign In or Register to comment.