View Single Post
Old 05-18-2017, 03:15 PM   #1
gruberjd
Member
 
Location: Michigan

Join Date: Nov 2011
Posts: 13
Question Calling variants when mapping reads from different species than reference

Hi all,

My SNP calling pipeline seems to be poorly performing when tasked with mapping reads from a sample that is from a related species from the reference assembly. (This sort of thing is common in plant breeding.) Obviously the more variants that differentiate reference and sample, the harder it is to map reads correctly, so it makes sense that this is harder than finding intraspecies SNPs.

Nonetheless, I am 100% confident it can do better than it is doing. I can see from the bam that there are reads mapping to a given region, and even then most of the time my SNP caller (the last free version of GATK) fails to emit genotypes, even of sites that simply match the reference. Interestingly, the few times that it does emit a genotype, it is highly highly biased towards A&T calls.

Is any of this a classic symptom of a parameter I need to adjust, or a sign that I need to switch SNP callers? Can anyone cite literature that compares SNP callers' performance (or influence of different parameters) in this particular challenging task? I haven't had much luck searching.

Thanks!

Jonathan
gruberjd is offline   Reply With Quote