I'm working on rather standard whole-exome sequencing data and treat it the same as whole-genome sequencing data (aligning it to the full GRCh38 reference assembly and calling variants with no exome-related parameters).

I recently noticed that my variant calling results contain numerous variants mapped to a non-exomic region, incidentally, some of them were significantly associated with the phenotype I study.

My questions are as follows:

  1. Are the non-exome variants in my exome data invalid? or could it be that the exome kit captured more than it was designed to capture?
  2. Should I treat raw exome sequencing data any different than genome sequencing data, and if so, what parts of my pipeline should be modified to accommodate exome data?

1 Answer 1


These are called off-targets due to unspecific binding of DNA to the capturing beads during library preparation. This is normal and expected since capturing is not a perfect process. It is correct to align against the whole genome but you should simply limit variant calling to exons (or basically the on-target regions) and ignore everything else.

  • $\begingroup$ Thanks for the reply, makes perfect sense. Now, considering one of my significantly associated variant is from an off-target region, should I report it as a legitimate direction for further research? Consider that the variant has adequate coverage, passes strict filtering in the entire cohort and heavily implicated in my phenotype of interest. $\endgroup$
    – shepp
    Jun 14, 2020 at 22:16
  • 2
    $\begingroup$ I would also call variants a few bases out of the exons to cover splice site mutations. $\endgroup$ Jun 15, 2020 at 7:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.