We've moved!
This site is now read-only. You can find our new documentation site and support forum for posting questions here.
Be sure to read our welcome blog!

GATK3.8 vs GATK4 HaplotypeCaller

Hello, maybe I'm asking a naive question and maybe it has been answered somewhere else, but as the title states are there differences in the algorithm of the HaplotypeCaller between GATK3.8 release and GATK4?
I'm calling variants in two exomes, using the same bam files as input of the variant caller, and I tried to compare variants at the end of the pipeline. I found that GATK4 calls a slightly higher number of variants than GATK3.8 and that this one in the final step step (after filtering and selecting variants) retains variants filtered away from the GATK4 HC, is this expected or do I have to go trough my pipeline and check for errors?
Thank you


  • SheilaSheila Broad InstituteMember, Broadie ✭✭✭✭✭


    I think there may be minor differences in edge cases (low quality or low coverage areas). But, the outputs should be functionally equivalent. What exactly is "slightly higher"? Do you have exact numbers? What kind of differences are there after filtering?


Sign In or Register to comment.