Hi,
I am performing VQSR (GATK 3.7) using the SNP model on individual chromosomes on hundreds of WGS data. However, for some chromosomes, it ran without a problem using the --maxGaussians default, which is 8. We followed your suggestion to lower that number to 4, it still doesn't work. We tried 3, still doesn't work. Finally it works when --maxGaussians is set to 2.
Apparently this --maxGaussians parameter has nothing to do with the chromosome size. My questions are
1) if --maxGaussians = 2 scientifically sound?
2) would you suggest using a uniform --maxGaussians parameter across all the chromosomes, i.e. 2 in this case, or I should just tune the parameter per chromosome?
Can you provide some guidance on this?
Thanks!