Filtering Non-Homozygous and Non-Heterozygous Variants in DNA Analysis Pipeline

Question

[ManuelDB]

We have a pipeline that analyzes somatic and constitutional human DNA for clinical diagnosis. The variant calling flags anything with a Variant Frequency (VaF) equal to or higher than 1%. We want to filter out variants that are neither homozygous nor heterozygous, as they are more likely to be called by artifacts or mosaics that we do not care about. Is there a method to do this?

This must be an old problem with a well-developed solution, but I cannot find it.

Answer 1

[Ram]

Unless I'm mistaken, variant calling should already remove all HOM-REF sites. If not, you can use bcftools view to exclude them:

bcftools view -e 'GT="ref"' -Oz in.vcf.gz -o out.vcf.gz