Homozygous SNPs in reference based consensus map-back

[AB.PATEL]

HI

I have performed reference based consensus generation. Next I have mapped back reads to consensus and performed SNP calling. Surprisingly I am getting ~15% homozygous SNPs. Is it possible to get homozygous SNPs when we are mapping reads back to consensus ?

[Brian Bushnell]

Perhaps you could go into a little more detail about your methodology. It's certainly possible, given that a new reference will affect mapping, and mapping can be nodeterministic anyway. Perhaps some areas were so full of differences to the reference that reads didn't map there at all, but now that the edge has been corrected, reads are mapping further in. Also, some variant-callers (like mpileup+bcftools) mask certain SNPs that are near indels, even if they are homozygous. Once the indel is out of the way, it will probably give correct variant calls. So, maybe you need to perform multiple cycles of refinement with your methods until the results converge.