How to Turn Multiple ALT alleles to Genotype Calls

Question

[jimkozubek]

I have a VCF file and many lines have multiple ALT alleles such as:

1       104159  .       CA      C,TA,TTT

I have an algorithm that takes genotype data in a 0,1,2 matrix form. I am wondering if there are any standards or best practices in how to turn VCF lines with multiple ALT alleles (0/0,0/1,0/2,1/2,2/3) into 0,1,2 genotype form.

[jimkozubek]

Thanks for the pro tip!

[obidobi]

did not work xd Can you share the solution if you have?

If I use the command below to divide them into biallelic, it does not work correctly.

Another solution is maybe to get rid of them but I do not lean towards.. Would be appreciated if you share the solution. Thanks!

Answer 1

[Kevin Blighe]

Yes, you can split multi-allelic calls with

bcftools norm -Ov -m-any MyVariants.vcf > MyVariantsSplit.vcf ;

A useful addition is to also set the variants such that the REF allele matches that of a chosen reference genome, for example:

bcftools norm -Ov -m-any MyVariants.vcf | bcftools norm -Ov -f human_g1k_v37.fasta > MyVariantsSplitRefChecked.vcf ;

Kevin