Number of missing genotypes in bcftools stats

[mufernando]

Hi,

How does bcftools stat compute number of missing genotypes per sample when presented with a "GATK" VCF (which encodes missingness in the DP field)?

Thanks,

Murillo

[pd3]

I am guessing your question is related to the bizarre GATK decision discussed also here samtools/hts-specs#756?

Missing genotypes are recognized by GT=. or GT=./., not by the DP or any other field. The assertion made in GT can come from various experiments where DP may not even be defined.

Note there are the -i/-e filtering options and the +setGT plugin (http://samtools.github.io/bcftools/howtos/plugin.setGT.html) which can be used to control which genotypes should be counted.