When to remove releated samples in GATK

Question

[Filago]

Hello Guys,

I am generating my VCF files with the GATK pipeline including joint calling GVCFs. The joint calling process calculates cross-sample information like QD, MQ... based on all samples in the GVCF. However afterwards when receiving the VCF files I (might) remove samples according to ancestry or relatedness. Thus the filters might be partially based on samples not present anymore in the data.

How do you solve this? Do you just apply 2 pipelines? One to remove samples and one to process the remaining ones?

Best,

Andreas