Dear all, please could you advise :
what is the best strategy to call the somatic variants in the absence of germline sample (when only the tumor sample is available) ? thank you,
-- bogdan
If your goal is to discover new somatic variants, I am afraid that might be quite challenging because the VAST majority of variants in any sample will actually be polymorphisms and not somatic variants. Use of a set of databases of "normal" variation to remove as many germline variants as possible can reduce the false positive results, but this approach will be clearly inferior to having a matched normal. If, on the other hand, the goal is to identify known cancer variants present in the samples being sequenced, having tumor only is not at all unusual and is a viable strategy.
hi, can you give a more detailed description about your last sentence “. If, on the other hand, the goal is to identify known cancer variants present in the samples being sequenced, having tumor only is not at all unusual and is a viable strategy.”， I want to know more, thanks a lot.
Are you looking for sensitive identification of all somatic variants, particular actionable variants (eg gene fusion breakpoints), or signs of large-scale events such as chromothripsis?
If you're happy to control your FDR by accepting a low sensitivity and ignoring small to medium sized indelsthen it's viable as I demonstrated here. If you're interesting in TE reactivation then you're going to need a normal as it's practically impossible to determine if those events are germline or not.
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
Dear Sean, great to hear from you, and a happy and fruitful spring time !
Talking about calling somatic variants without matched normals, I came across VarDict from AstraZeneca : https://github.com/AstraZeneca-NGS/VarDict. If anyone has any feedback about VarDict, please let me know ;)