Clipping reads and reference bias

[danchubb]

Hi,

We have noticed that the Illumina aligner Isaac soft clips the end of reads that have non-reference bases https://www.biorxiv.org/content/10.1101/836171v1. This affects battenberg results so to correct it we have forked a version of allelecounter that removes biased counts by clipping all reads by n bases (e.g. -e 5 end clips 5 bases) https://github.com/danchubb/alleleCount-FixVAF

The main change is in https://github.com/danchubb/alleleCount-FixVAF/blob/dev/c/src/bam_access.c line 131 where the count is skipped if a position is within e bases of the end.

It might be something you'd want to merge in or at least be aware of if you use allelecounter for isaac aligned data.

also, this seemed to be the easiest way of dealing with it in allele counter and I don't think it will break anything else but please let me know if you have a more elegant solution.

Thanks,

Dan

[keiranmraine]

At first this didn't make sense being that soft-clipped are already excluded, however the article makes it more clear.

I don't see anything wrong with the method applied, however raising a pull request would be the best way to proceed... @drjsanger is the gate keeper here

Thank you for contributing!

[keiranmraine]

Specifically section 2.3:

The Isaac --clip-semialigned parameter invokes the soft clipping of read ends until five consecutive bases are matched with the reference genome

[ghost]

Looks like a very reasonable addition. As per @keiranmraine suggestion. Please make a pull request, we'll merge it into our codebase. Thank you for contributing @danchubb !