Can i use tumor samples to generate reference files?

[weizhiting]

Hi, thanks for this wonderful tool. I have bam files of two groups patients, A group and B group. Every patient has normal sample and tumor sample. I want to get genes that are enriched in B group compared to A group. One way to do this is: compare A group tumor to A group normal, compare B group tumor to B group normal. and use the segment file to run gistic2, separately. Finally, genes enriched in B and not in A is the results. Another way is to use A group tumor sample as reference directly, and compared it with B group tumor sample. then run gistic2 with the segment file. So i want to ask is this method reasonable? Thanks for your reply!

[philtat]

Yes, the -n option will do this

https://cnvkit.readthedocs.io/en/stable/quickstart.html#build-a-reference-from-normal-samples-and-infer-tumor-copy-ratios

from the link above: "If you have no normal samples to use for the reference, you can create a “flat” reference which assumes equal coverage in all bins by using the --normal/-n flag without specifying any additional BAM files:"

[weizhiting]

Sorry for that i did not make it clear. I want to make sure if can i use the command cnvkit.py reference *Tumor.coverage.cnn -f ref.fasta -o Reference.cnn to build reference, which Tumor.coverage.cnn is the tumor sample coverage.

[etal]

Yes: giving these tumor samples as input to the reference command will work the same as giving these samples to batch -n.