combine target and exome data?

I have a cohort of samples with some data including whole exome and others limited to a targeted panel of ~50 genes. Would it be problematic to generate gVCFs for all the samples using the appropriate intervals and then combine them with GenotypeGVCFs? Since the libraries are prepared in different manners I was not sure if this would create problems during the variant calling step.


  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin

    Hi there, it's not problematic for the variant calling step, but it could be a problem in the filtering step. The VQSR model may be perturbed by heterogeneity in the variant context profiles -- if they are different between your exomes and gene panels. You can test this fairly easily -- once you have GVCFs for all samples, try running the following steps on the whole cohort vs subsets by experimental type, and see how much difference you get. And let us know what you find, as it could be of interest to others.

Sign In or Register to comment.