Forum:Multisample SNP/SNV calling - What's your favourite method?
1
0
Entering edit mode
6.3 years ago
amcrisan ▴ 350

I have been using the multi-sample variant calling (SNP or SNV) calling method that's a part of samtools and I am not entirely certain how confident I am in it's output. The feeling is subjective - when I run each of the sample files individually and look for shared variants and differences between the samples I get a set of variants that I am fairly confident in. They have good quality, are shared by multiple samples (but not all samples), and they seem to be fairly robust (i.e. show up again and again) in spite of the changing the strictness of mapping or base quality criteria. Some of those pretty robust variants are absent from the final filtered VCF when I run the multi-sample variant calling method.  There was one variant in particular that was present in all but one sample that did not get picked up by the multi-sample variant calling samtools method. I have a lot of samples and the average coverage across the genome, per sample, is quite high (~100 ; I removed the -D filter).  I know the multi-sample method in samtools was optimized to run on the lower coverage data from the 1000 genome project, so perhaps part of the discrepancies stem from underlying statistical machinery being tuned for such data.

With this is mind I wanted to get a sense of what people are using for multi-sample SNP/SNV calling; whether a lot of people do use it (as opposed to calling variants on each sample separately then amalgamating the information using a script after the fact); and whether the results of samtools SNP/SNV calling, specifically, are reliable (in theory, the multi sample method should be more reliable than the single).

 

 

SNP sequencing Forum • 2.6k views
ADD COMMENT
0
Entering edit mode
6.3 years ago

I haven't really worked with pooled data (if that's what you mean by multi-sample SNP calling) but I've been using SNVer, and it works great. One thing I really like about it is that it lets you define the ploidy of your sample, which is helpful for me since I work with haploid yeast data a lot. Samtools assumes that you're using diploid samples, so if you're looking for SNVs, it won't be accurate. SNVer also supports pooled data.

ADD COMMENT

Login before adding your answer.

Traffic: 1275 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6