To celebrate the release of GATK 4.0, we are giving away free credits for running the GATK4 Best Practices pipelines in FireCloud, our secure online analysis portal. It’s first come first serve, so sign up now to claim your free credits worth $250. Sponsored by Google Cloud. Learn more at

Recommendation on performance using scatter/gather

LaurentLaurent Member, Broadie
edited April 2013 in Ask the GATK team

Dear all,

I am currently running an analysis using the HaplotypeCaller on 300 large BAM files on our cluster and decided to chunk the the genome in 3MB bins in order for them to be processed in a decent time. I'm however experiencing very long runtimes as more and more jobs get scheduled to run in parallel on the same files. Looking at the GATK options, I saw these 2 that I thought could be of help and was wondering what were the recommendation for using them:

More precisely, does the num_bam_file_handles increase processing time by a lot? and what is the default value for --read_buffer_size ?

Thanks a lot,


Best Answer


  • LaurentLaurent Member, Broadie

    Hi Mark,

    Thanks for the explanation! I will digg more into the problem, but at the moment what I am reporting is only "observation" of my runtimes getting extremely high when running in parallel. I have observed similar problems when running other walkers using scatter/gather in the past on our cluster. I'll give a shot at extracting the regions using PrintReads to a local scratch beforehand and let you know if this helps! I'll also use the nightly build to benefit from the latest improvements.

Sign In or Register to comment.