What are the steps for Germline and Somatic CNV detection using CNVkit?

Question

[ahmad mousavi]

Hello,

I have some basic questions regarding the CNV calling using CNVkit. I have tried to find the CNV caused between normal and tumor samples, but it was not clear which steps should be done respectively.

1. For WGS (Somatic) tumor and normal samples, Should I first run the CNVkit batch then CNVkit call ? or "batch" is enough?
2. What would be the WGS Germline CNV calling steps?
3. How can I find filter the founded CNV, I mean the criteria for cnr or cnn ? ( I mean the significant CNV for the genes).

Thanks a lot.

[Ram]

Just from reading the manual, it is clear that batch gives you log2 CN values and call gives you thresholded integer calls from the log2 values. We cannot tell you what you should do - you need to use the commands to get you to the endpoint you need.

[ahmad mousavi]

thanks a lot Ram for the answer. If I want to clarify it, we would like to search the CNV that cause the cancer, both Germline level and Somatic level.

I read the manual, but it was not clear is "batch" is good? which threshold is good for the cancer related studies ? log2 >=2 would be enough?

Thanks a lot.

[Ram]

it was not clear is "batch" is good?

I'm not saying that the manual says batch is "good". Good and bad are subjective to where you need to get to. I said batch gives you log2 CN values. Is that what you need? Then batch is "good". If not, you need something more.

I don't know a lot about CN values, but I'd recommend you read papers on the topic. Pick a recommended threshold or a threshold used by a well-known, well-reviewed paper or something that you can explain. And please understand that at best, you can show correlation; you cannot prove a CN causes cancer through informatics alone.

[ahmad mousavi]

thanks Ram for the answer.

[ahmad mousavi]

Dear Ram,

How can I filter the genes from output.call.cns ? Should I filter all row with CN=2 ?

[Ram]

I don't know about that.