Sundial's final.full.vcf which is a proxy for clockwork's gvcf doesn't use the same fields. Gvcf lacks a COV field, instead using DP for this. Sundial, being more sane, uses DP to mean total depth and has a COV field already.
Before this, sundial's DP field was overriding with COV, and pysam's parsing treated both as the first value (which didn't have expected values)
This didn't cause an error though as this is technically fine and a valid VCF. It did however cause issues around merged in rows with expected minor populations (which rely on the COV field to be found)
Sundial's
final.full.vcf
which is a proxy for clockwork'sgvcf
doesn't use the same fields. Gvcf lacks a COV field, instead using DP for this. Sundial, being more sane, uses DP to mean total depth and has a COV field already. Before this, sundial's DP field was overriding with COV, andpysam
's parsing treated both as the first value (which didn't have expected values)This didn't cause an error though as this is technically fine and a valid VCF. It did however cause issues around merged in rows with expected minor populations (which rely on the COV field to be found)