How to tell whether indel calls in NGS are artefacts caused by homopolymer (e.g. poly-A) regions

Question

[CMM]

Hi,

I am relatively new at working with NGS data (i.e. I am familiar with the basic concepts of QC, alignment and variant calling, but have less experience with downstream analysis of the data).

I have hybrid-capture sequencing data of human cancer tumour cells and I would like to check whether the indels we are calling are "real" or whether they are caused by homopolymer regions (which I have heard can be a common problem).

My approach would be to identify the insertions/deletions in the VCF file and then manually match them against the reference genome to see whether there are polyA or polyT regions there.

Is this the correct approach? Or can someone suggest a better tactic.

Thanks in advance.

Answer 1

[WouterDeCoster]

We annotate our variant files with repeat tracks, including homopolymers, and (most often) discard variants in the homopolymers (since they have a high likelihood of being false positive). You can do this with snpeff (http://snpeff.sourceforge.net/SnpEff_manual.html, under the title "adding your own annotations")

[CMM]

Thanks Wouter, that sounds like a good solution! I'll give it a go.

Answer 2

[WouterDeCoster]

I just came across this blog post as another potential solution: http://apol1.blogspot.be/2014/05/how-do-i-identify-all-homopolymer-run.html

Surprisingly fast!