Why is Clair3 run in the run_deepsomatic workflow?

[TimD1]

Firstly, thanks for sharing this WDL pipeline, it was exactly what I needed!

I'm looking for clarification as to why Clair3 is run twice (once to call tumor germline variants, and once to call normal germline variants) in the run_deepsomatic workflow. Shouldn't these germline callsets be the same (ignoring possible mosaicism)? It looks like only one is used downstream.

DeepSomatic is used to call the somatic tumor variants (filtering out the germline tumor variants). Why not keep both outputs from DeepSomatic, and skip Clair3 altogether? Just curious if there were reasons that this wasn't done (accuracy, etc). Thanks!

[proteinosome]

Hi Tim.

Thanks for opening the issue.

Clair3 is ran on both because in some experimental designs the normal coverage can be very low, and so calling germline variants on the normal can miss a lot of variants. Calling it on the tumor, while not perfect, leaves the options open if users want to look for specific germline variants that's missed from shallow coverage in the normal. Of course, you will never be sure if they're really germline, or somatic, but I think it could still be useful.

Germline variants are useful in the case of for example BRCA variants that associates with onset of certain types of cancer.

Hope this answers your questions, let me know if you have any other questions!

Side note: I'm working on a new version of the pipeline and so expect changes coming in the next one month or so, e.g. new versions of tools and better/more robust reporting.

Thanks.