How to make a representative copy number variation sample from several samples based on their overlaps?

Question

[nazaninhoseinkhan]

Dear all,

I want to compare copy number variations between a tumor and its corresponding normal samples for all my tumor samples.

I am working on TCGA data and unfortunately all tumor samples have not their corresponding normal samples.

I want to make a representative copy number variation samples from several normal samples and then compare each individual tumor sample with that artificial normal sample.

Can you explain me the best strategy for this task?

Should I use GenomicRanges package?

I have the segment_mean for different ranges in each chromosome and I do not have access to raw data.

Thank you in advance

Nazanin

Answer 1

[jared.andrews07]

If you had the raw data (assuming it's WGS or exome), I'd recommend CNVkit, as it can create a reference normal sample for you to which it'll compare each of the tumors.

Since you only have segmentation data, maybe GISTIC2 will work. You can provide it a list of "germline" CNAs (i.e. those from your normals), that it will exclude from the analysis of the tumors.