HalplotypeCaller and genotypeGVCFs in joint calling
HC has four major steps: 1) Define active regions. 2) Determine haplotypes by re-assembly of the active region; 3) Determine likelihoods of the haplotypes given the read data; 4) Assign sample genotypes.
In joint calling, which of those steps 1-3 uses the pulled reads from the cohort instead of run within each sample like step 4?
If any of the first three steps uses cohort reads, how is it done if all sample files are in gVCF?