how to normalize and compare CNV value in different samples

Question

[Shaway]

Hello everyone,

I have two groups of samples, tumor and normal. Now, I want to compare the CNV difference between this two groups.

I windowed the genome in to 10kb bins, and compute the reads number in every bins. I have to normalize the sequence depth and make sure the CNV value in every bin are in the same scale in order to have a valid comparison.

I used HMMcopy to tackle GC bias, but how to scale the different samples into a comparable dimension puzzled me. I know that just normalize sequence depth is not enough.

Is there any software or methods to tackle this problem?

Sorry for not illustrate this in picture, I failed to insert pictures in this post.

Thanks!

Shaway

Answer 1

[jared.andrews07]

You could try CNVkit. It uses your normals to create a reference to which it'll compare each of your tumors. It's pretty quick and easy to use, so might at least be worth a shot for your case. Certainly easier than trying to do all the normalization by hand.

[Shaway]

Thanks for your help!