Hi Ilya,
The JVM can be quite a pain when determining the best way to determine max
memory, however the underlying gatk_wrapper.py script was updated last week in
5970:20215fcf6da7 to allow using -Xmx or XX:DefaultMaxRAMFraction. Each of the
GATK tools are currently already set to use 'XX:DefaultMaxRAMFraction' of 2
(half of ram) - future enhancements to the tool running code will allow this to
be set based upon e.g. number of jobs and available ram per node (no time frame
for this).
It does look like num_threads/nt is now available for the UnifiedGenotyper.
Many of the GATK tools list it as an available parameter, but when run will
give an error saying it is not yet implemented (only can determine by trial and
error or reading source code). I've updated the Unified Genotyper to use
--num_threads 4. Thanks!
It might be a good idea to coordinate the changes you are looking into with
GATK with the changes we are approaching. Any other feedback would also be
greatly appreciated.
Thanks for using Galaxy,
Dan
On Sep 7, 2011, at 1:25 PM, Chorny, Ilya wrote:
> Hi Daniel,
>
> I have been implementing the GATK unified genotyper and I was having some
> issues. Occasionally I would get an error that the Java virtual machine would
> not start. I got around that by adding the Xmx3g parameter to the command
> line. I also added the –nt 3 because otherwise it’s really slow on human
> data.
>
> --stdout "${output_log}"
> #for $i, $input_bam in enumerate( $reference_source.input_bams ):
> -d "-I" "${input_bam.input_bam}" "${input_bam.input_bam.ext}"
> "gatk_input_${i}"
> -d "" "${input_bam.input_bam.metadata.bam_index}" "bam_index"
> "gatk_input_${i}" ##hardcode galaxy ext type as bam_index
> #end for
> -p 'java -Xmx3g
> -jar "${GALAXY_DATA_INDEX_DIR}/shared/jars/gatk/GenomeAnalysisTK.jar"
> -T "UnifiedGenotyper"
> -o "${output_vcf}"
> ##-o "out_vcf.txt"
> -et "NO_ET" ##ET no phone home
> ##-log "${output_log}" ##don't use this to log to file, instead directly
> capture stdout
> #if $reference_source.reference_source_selector != "history":
> -R "${reference_source.ref_file.fields.path}"
> #end if
> -nt 3
> --standard_min_confidence_threshold_for_calling
> "${standard_min_confidence_threshold_for_calling}"
> --standard_min_confidence_threshold_for_emitting
> "${standard_min_confidence_threshold_for_emitting}"
>
> Ilya Chorny Ph.D.
> Bioinformatics Scientist I
> Illumina, Inc.
> 9885 Towne Centre Drive
> San Diego, CA 92121
> Work: 858.202.4582
> Email: [email protected]
> Website: www.illumina.com
>
>
>
>
___________________________________________________________
Please keep all replies on the list by using "reply all"
in your mail client. To manage your subscriptions to this
and other Galaxy lists, please use the interface at:
http://lists.bx.psu.edu/
