SYNOPSIS

blt vcf-downsample desired-count < input.vcf > output.vcf

DESCRIPTION

blt vcf-downsample Outputs a uniformly random sample of roughly desired-count VCF calls from the input. The strategy involves generating a random number which is compared to desired-count / original-count for each input VCF call, and outputting it only if the random number is less. As a result, the actual number of output calls is usually not exactly desired-count, but is usually within a few percent.

The program is quite fast, so if the actual count is too far off for your needs, one strategy is to simply discard the output and try again.

Currently only single-sample VCFs are supported. Multi-sample VCFs will be supported if and when a need is encountered.

EXAMPLES

blt vcf-downsample 1000 < input.vcf > output.vcf

SEE ALSO

blt(1)

AUTHOR

J. Bacon