Paired Tumor/Normal workflow for ASCAT sc; sample concordance analyses for sWGS

[ahwanpandey]

Description of feature

Hello,

Would there be a possiblity of running ASCAT sc through the SAMURAI pipe and providing a matched germline/normal BAM?

https://github.com/VanLoo-lab/ASCAT.sc/blob/af9add3305582310b7284a76d198eeb6e1da6288/R/run_sc_sequencing.R#L17C31-L17C42

Also a question for you, are there any tools for sWGS that can check sample-to-sample concordance / swaps / contamination like the following tools can for other types of data? I don't suppose these tools would work for really low coverage <1x sWGS data? Just asking as I don't have much experience doing sample concordance analyses on such shallow data.

https://github.com/PapenfussLab/HaveYouSwappedYourSamples
https://bamixchecker.readthedocs.io/en/latest/

Thanks!
Ahwan

[lbeltrame]

Hello!

We use PoNs exclusively but I think it's a valid feature request. It needs a new parameter and some wrangling logic in the ASCAT.sc script, but it is certainly doable.

Also a question for you, are there any tools for sWGS that can check sample-to-sample concordance / swaps / contamination like the following tools can for other types of data? I don't suppose these tools would work for really low coverage <1x sWGS data? Just asking as I don't have much experience doing sample concordance analyses on such shallow data.

We've been thinking about this too. So far the only idea that came up was to use known dbSNP or gnomAD loci, where technically you already know what the expected alleles are, perhaps using this project, but then we'd need to make a VCF or something ourselves and that's where it gets tricky...

[ahwanpandey]

Ok thanks re: matched germline/tumor analysis.

Re: Sample concordance analysis, the tools I mentioned and the tool you pointed to are basically all doing the same thing. They work really well with the standard coverage data. But they dont work as well with shallow WGS data, at least the data I am looking at which is ~1x. Which makes sense becuase we need enough coverage to call SNPs with confidennce. I just wanted to ask if you knew of other ways as I don't have much experience analysing such shallow coverage WGS. But seems it is a limitation of extremely shallow coverage data.

[lbeltrame]

As I said, a possibility would be to count alleles directly on known polymorphic sites (so no need to call) and extract a crude allelic ratio. But I fear this might be very fragile.
ASCAT.sc uses a similar approach to call allele-specific CN using sWGS (I didn't test it personally, but I know people who did).