Running CNVkit

[eesiribloom]

Thanks for putting CoRAL out, Im excited to try it and see how it compares to decoil and AA. What parameters were used in the paper to run CNVkit as it is not strictly "designed" for long-reads? If I already have CN calls from CNVkit in matched short-read data would that be suitable?
Thanks

[jluebeck]

I think this should be fine - but with the caveat that the short-reads should be truly matched and from the same biospecimen (not generated from a sample taken weeks or months earlier/later, or taken from a different region of the tumor, etc.). It is a bit alarming how quickly even cancer cell lines seem to evolve.

There are also some other options for long-read CNV calling emerging (e.g. Spectre) that may be good alternatives. I have changes pending in an open PR on CoRAL to enable use of other tools' .bed files.

Thanks,
Jens

[eesiribloom]

Okay thanks. Would you recommend long-read specific CNV caller e.g. spectre or continue using CNVkit ?

[MJoseMo]

Hi, did you manage to implement spectre and run CoRAL?? @eesiribloom

[eesiribloom]

@MJoseMo I actually just used CNVkit with long reads. Seems to work okay but at some point I will probably compare results with spectre and can let you know