This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Removing alternate alleles from .vcf

I had whole genome sequence data in .vcf format from several different individuals. I extracted a SNP set from each individual, removed any SNPs with more than 2 alleles, and then merged them all together in bcftools.

Everything seems OK, other than there are several sites which have more than 1 alternate allele in the merged dataset. For example:

1       776546  .       A       G,T,C   287     .       GG=257,297,0,297,730,285,730,221,285,730;DP=135 GT:PL   0/1:257,0,221   0/1:86,0,133    0/0:0,12,165    0/1:255,0,325   0/1:337,0,77    0/0:0,3,46      0/0:0,129,1000  0/0:0,42,291

However, if you notice the genotypes, they are all either 0/1 or 0/0, i.e. there are only 2 alleles present in the callset. The excessive alternate alleles is messing up a new merge that I want to do, because bcftools is saying that there are 4 alleles, but only 3 PL score entries.

Does anyone know of a way to trim the vales in the ALT column on the vcf, so that there is the 'correct' number, given the number of different genotypes?

EDIT: Ive just found the command bcftools view --trim-alt-alleles. I think it's done what I hoped it has, but the documentation isnt very descriptive. Could someone confirm what it does? Thanks.

next-gen
• 2,626 views • link
updated by zjardyn • zjardyn • written by ucbtsm8 • ucbtsm8

1 answer

In R you can check if the .vcf contains alternate alleles with the following code:

library(vcfR)

vcf <- read.vcfR("example.vcf")

gt <- extract.gt(vcf)

unique(unlist(apply(gt, 1, unique)))

[1] "0/0" "1/1" "0/1" "2/2" "0/2" "1/2" "2/1"

In this case 0 is the reference, 1 is the first alternate and 2 is the second alternate.

• 0 views • link
written by zjardyn • zjardyn

Log in to answer this question.

More posts like this