Number of samples and variants in .vcf file

Question

[bluebubbl19]

How can I count the number of variants (lines) and samples in a vcf file using simple bash commands (not using vcftools, bcftools, gatk or another packages)?

Can I use wc -l for variants? How can I get rid of the lines which do not represent variants? Is there any similar wc command that would give me the number of samples?

Each of my chromosomes is one file.

Thank you!

Answer 1

[4galaxy77]

The number of variants is given by

zcat in.vcf.gz | grep -v '#' | wc -l

The number of samples is non trivial if it doesn't contain genotypes, but if it does then try:

zcat in.vcf.gz | grep -v '#' | head -n1 | sed 's/\t/\n/g' | grep '/' | wc -l

That said, I can't imagine any reasons why you wouldn't just use bcftools stats which is way more robust, simpler, and gives you a ton more information. Don't resort to regex tricks when you can use a dedicated tool like bcftools.

Answer 2

[Pierre Lindenbaum]

number of variants:

zcat in.vcf.gz |  grep -c '^[^#]'

number of samples:

zcat in.vcf.gz | grep -m1 "^#CHROM" | cut -f 10- | tr "\t" "\n"  | wc -l