exome CNV: remove duplicate reads?

[mrfox]

Hi all,

I just want to know if you remove the duplicate reads when detecting CNVs using whole exome sequencing data. It seems that for SNV calling duplicate removal is widely accepted. How about CNV detection?

Thank you.

[fromer]

Hi,

Yes, PCR amplification distorts the signal of expected correlation between underlying copy number and read depth. Therefore, it makes sense to try and remove the reads that are clearly duplicates and artifacts of the PCR. Of course, there may be more subtle sample-specific and target-specific artifacts (or reads not marked as duplicates, depending on the stringency of the marking), so it's still necessary to "clean up" the data to detect copy number.

We describe in our recent paper our approach to doing this, which includes both data-driven normalization using PCA (principal component analysis) and using an HMM (hidden Markov model) to statistically genotype CNV:
Discovery and statistical genotyping of copy-number variation from whole-exome sequencing depth. American Journal of Human Genetics, 91:597-607, Oct 2012.

http://www.cell.com/AJHG/abstract/S0002-9297%2812%2900417-X

And, XHMM, our downloadable software for detecting CNV from exome data can be found here:

http://atgu.mgh.harvard.edu/xhmm/

Best,
Menachem

[mrfox]

Thank you very much Menachem. It is great to have published papers to support the duplicate removal. And thanks for the software. I will try it.