filter unique variants from multiple VCF files

Question

[anu.pullanhi]

Hello

I have multiple normal and benign VCF files. I want to filter unique variants between normal and benign. How can i perform the filtering?

Thanks in advance.

[Kevin Blighe]

I want to filter unique variants between normal and benign

So, you want to find variants that are called in at least 1 benign sample that are not called in any normal sample (and vice-versa)?

[anu.pullanhi]

Yes, filter unique variants that are called in at least one benign sample that are not called in any normal sample and vice versa.

Answer 1

[finswimmer]

Before starting you have to merge all your vcf files into one, with e.g. bcftools merge.

After that, you can use bcftools view to find your variants of interest. Create a file with the sample names for your "normal" and one for your "benign" (one sample name per line). Than try this:

bcftools view -i '(COUNT(GT[@normal.txt] = "alt")> 0 && COUNT(GT[@benign.txt] = "alt") = 0) || (COUNT(GT[@normal.txt] = "alt") = 0 && COUNT(GT[@benign.txt] = "alt") > 0)' merged.vcf.gz

(Couldn't test it for now)

Make sure you are using a current version of bcftools (v1.9).

fin swimmer

[anu.pullanhi]

Hello

While using bcftools merge, i get error "Not ready for type [0]: SNP at 12783".

[finswimmer]

Hello again,

could you please show the complete header (the lines starting with a #) and the first few variant lines of your vcf file?

fin swimmer

[anu.pullanhi]

sorry! i cannot share the data.

Thanks

[finswimmer]

Without looking at your data it is quite hard to help. Maybe this helpful for you: Is there a way to Scan and Correct bad vcf INFO values?

[anu.pullanhi]

I understand that, but i cannot share the data. I already looked into the issue and I don't find any AS_MQ error in the VCF file. I also checked SNP as the error was showing as "Not ready for type [0]: SNP at 12783".

##INFO=<ID=SNP,Number=A,Type=Flag,Description="Variant is a SNP">

[Kevin Blighe]

I have just encountered this same error message.

This problem is elaborated here: https://github.com/samtools/bcftools/issues/1685

You need to identify tags like this:

bcftools view -h "${vcf}" | grep -e "Number=A,Type=Flag"

Then, either follow the advice in the GitHub issue and recode them as Number=0,Type=Flag, or remove these if they are not needed:

bcftools annotate -x INFO/SNP,INFO/MNP "${vcf}"