One usage example is to refine genotypes based on the discovered allele frequency in an input VCF containing many samples

 gatk CalculateGenotypePosteriors 
   -V multisample_input.vcf.gz 
   -O output.vcf.gz

Now if I have a case-control cohort of autoimmune disease, I wonder if I should run the above step separately for case and control, or put them into one vcf file?

Thanks for any insight!