Count missing and non missing values in a VCF file

Question

[LDT]

Dear all,

I have a vcf with 100s of populations, and I want to count how many missing and non-missing values I have in each sample, and then for each variant for all samples.

Does anyone have a tip on how to calculate these stats?

thank you for your time

[4galaxy77]

how to calculate %of missing genotype in vcf file - much easier to use plink2 --missing

Answer 1

[iraun]

Maybe How to calculate the number of SNPs in each sample in a multi vcf file helps?

For the missing variants, I have created this simple awk script to count the number of ./. genotypes per sample, let me know if it works.

NR==1 {
    for (i=10; i<=NF; i++) {
        f[i] = $i
    }
}
{
    for(i=10; i<=NF; i++) {
        if ($i ~ /\.\/\./){
            b[i]++}}
    }
END {
    for(i=10; i<=NF; i++)
        printf "%s %s\n", f[i], b[i]
}

Save the previous code in a file named sc.awk (for example), and run it using:

awk -f sc.awk multi_sample.vcf

[LDT]

Yes, Iraun. That's great! Any ideas on how to calculate the missing values for each variant? thank so much for your time

[iraun]

I have edited my question with a potential solution for the missing genotypes per sample (you said in your previous comment "for each variant", but I think you mean for each sample, right?

Btw the script is adaptable, if you replace /\.\/\./ by /1\/1/, it will count the number of homozygous mutations, 0/1 for the heterozygous, etc.

Answer 2

[raphael.B]

I think bcftools stats -S samples.txt file.vcf provides this kind of information among others