Bug Bulletin: The recent 3.2 release fixes many issues. If you run into a problem, please try the latest version before posting a bug report, as your problem may already have been solved.

Why is BaseRecalibrator so slow - not using all threads?

drchriscoledrchriscole Posts: 16Member

Hi,

I'm running BaseRecalibrator on a full lane of HiSeq. I set the -nct switch to 8, but I see that the average load is <1. Why is this so? Doesn't the BaseRecalibrator use threads properly?

I've seen this thread http://gatkforums.broadinstitute.org/discussion/1263/baserecalibrator-trade-off-between-runtime-and-accuracy on how to speed up the process by downsampling. Is this a better option considering I have >200M reads?

BTW I'm using version 2.2-8 - do newer versions implement -num_threads?

I would really love to reduce the runtime from 36hrs.

Many thanks.

Answers

  • drchriscoledrchriscole Posts: 16Member

    After a bit of investigating of parameters I've used the -dfrac 0.05 and -dt ALL_READS options to get my runtime down to ~3hrs.

    However, the low amount of CPU load is still a concern.

  • Geraldine_VdAuweraGeraldine_VdAuwera Posts: 6,080Administrator, GATK Developer admin

    You should definitely use the latest version to take advantage of recent improvements that have been made at many levels of the GATK -- such as the revamped downsampling engine which is less biased than the previous implementation.

    BaseRecalibrator does not support the use of data threads (-nt), only cpu threads (-nct) which you're already using. I can't say why you would see a smaller cpu load than expected; the tool itself uses cpu threads just fine in our hands. Some people have platform or filesystem-related issues getting multithreading to work properly, but that's not something we can do anything about at the moment.

    Geraldine Van der Auwera, PhD

  • Mark_DePristoMark_DePristo Posts: 153Administrator, GATK Developer admin

    Definitely use the latest GATK. But it's possible that your system's IO cannot handle the input / output requirements in some way, as you can quickly bottleneck on getting the data or writing it out for 8x throughout. Does the problem persist with the lastest GATK and with a local BAM file?

    -- Mark A. DePristo, Ph.D. Co-Director, Medical and Population Genetics Broad Institute of MIT and Harvard

  • drchriscoledrchriscole Posts: 16Member

    Thanks for the replies.

    I've monitored the process with top and I can't see any I/O waiting - it flicks between running and idling - but that might be too blunt a tool?

    I'll try a newer version and see how things progress.

Sign In or Register to comment.