How Large Can A Coverage Discrepancy Be Between Two Aligners, Specifically Bwa And Novoalign

Question

[Matthieu Miossec]

I am involved in a project where variants have been called using two different aligner/variant caller pipelines on the same illumina sequenced data and we have found that there are some major discrepancies in reported coverage for some of the positions in which variants have been found. While I expect there to be some difference in coverage between two aligners, am I right to suspect that 50+ difference in coverage is uncommon or simply not possible? The aligners used are well established ones: BWA and Novoalign.

[Fabio Marroni]

You should check (and if you wish, post) alignment parameters. Maybe one of the two software was used with stricter parameters? In my experience, the two aligners usually give similar results. Also, check your average coverage : if your coverage is 1000 becaues you are doing resequencing of a small target then a 50+ difference is not so great.

Answer 1

[Jeremy Leipzig]

Do your total counts resemble the ones below?

[Matthieu Miossec]

Thanks, I'll check.