GATK 4.beta.2 vs. 4.beta.3 performance
Hi, I've been keeping up with the GATK 4.0 beta releases and I've noticed some performance differences between my runs of version 4.beta.2 and 4.beta.3. That is, 4.beta.3 tools seem to have a longer execution time. The runtimes and tools in question are:
|BaseRecalibrator||1m 3s||3m 3s|
|ApplyBQSR (scattered)||4m 48s||11m 51s|
|HaplotypeCaller (scattered)||23m 42s||29m 7s|
|GenotypeGVCFs (scattered)||4m 6s||9m 28s|
|VariantRecalibrator (for SNPs)||4m 7s||6m 38s|
|VariantRecalibrator (for INDELs)||2m 7s||4m 8s|
|ApplyVQSR (for SNPs)||37s||2m 36s|
|ApplyVQSR (for INDELs)||39s||2m 35s|
Note: The scattered tools are not the Spark version, but are just scattered on intervals from a custom BED file. The FASTQs used for these runs are synthetic exome files (just for quick and easy testing purposes). The computation was done in cloud, not on a local machine. The roughly same runtimes were acquired after reruns.
I haven't found anything in the release notes that would suggest these results. Are these runtimes to be expected?
Thank you for your help!