VCFParseError: ploidy > 2 not supported

Question

[always_learning]

I have few SNP with ploidy > 2 in my VCF(Human) generated by GATK with genotypes like "0/1/1". Any idea how I can remove them from my VCF file? Any tools that do that?

[finswimmer]

Hello,

why have you run your variant calling with a parameter that produces ploidy>2? Can you show us please the complete command?

Do your really want to remove those sites from your vcf or should the genotype be fixed in some way?

fin swimmer

[always_learning]

I can't run this again. This is merged VCF of around 6000 samples. I think It will fine to remove such sites from VCF.

[finswimmer]

Should the site get removed even only one sample have a ploidy>2 or should the genotype get set to unknown?

[always_learning]

I think "genotype get set to unknown" will be a better approach. What do you think?

[finswimmer]

Try this sed command:

$ sed -r 's:\t[0-9]+/[0-9]+/[0-9]+[^:\t]*:\t./.:g' input.vcf > output.vcf

It will look for a tab followed by at least one number, followed by /, followed by at least one number followed by a / followed by at least one number, followed by anything else but a : or a tab and replace these pattern by a ./. preceded by a tab.

fin swimmer

Answer 1

[tpoterba]

This is a Hail (0.2) error message from VCF import.

While we don't support triploid calls yet, we did add an argument to import_vcf: filter='\t0/1/1' will remove all lines matching that regex before parsing.

If you want to keep these sites but mark these calls as no call, then you can use the find_replace argument as well: find_replace=('\t\d/\d/\d', './.')