mutation detection in tumors samples

Question

[Bogdan]

Dear all,

we have a few samples where only tumor cells have been sequenced (no known normal pair); please could you advise on a mutation detection algorithm that could work well only on the TUMOR sample. Many thanks,

bogdan

Answer 1

[poisonAlien]

Without matched normals its hard to decide whether the detected variant is a true somatic event or not. People do all sort of filtering and comparision to exclude possible germline events. Check out is thread for details.

I think MuTect can still run without matched normal samples, but no matter how sophisticated the algorithms are, without germline control there is always uncertainty.

Answer 2

[Chris Miller]

This has been covered pretty extensively on the site before - here's one example (and there are others). Calling variants in blood tumor samples without matched normal samples

Any single-sample caller will generally work, as long as it doesn't have expectations about the VAF of the variants (since in tumors, they will not necessarily be at 50/100%).