Here are 2 scenarios involving 3 WGS samples with depth of coverage 5X-10X:
A- Feed the 3 bam files into HaplotypeCaller and run in discovery mode;
B- convert each bam file into a GVCF, and then perform joint genotyping on the 3 GVCFs.
Questions:
1- Will scenario A produce more accurate calls because the 3 samples are 5X-10X. If the difference in accuracy is not that great, then I prefer to use scenario B because it is best practice.
2- In terms of total computer run time are both scenarios sort of similar, or would it be likely that one scenario is much faster