GATK variant calling for multiple exome

sehrrot · 04-23-2012, 06:38 PM

Hi all,

Currently, I am running Unified genotyper for variant calling for around 100 exomes. It's on the cluster and I found the remaining running time is aournd 3 weeks. Is there any way that I can do within few hours? It's my first time to analyze high-throughput data..

Any advice would be appreciated

adaptivegenome · 04-23-2012, 10:58 PM

That seems like a long time. Can you share some more details about what you are doing so we can see if there is opportunity for speeding it up? Certainly a generic answer would be to leverage a cloud service to use as much CPU power as you need but I would be interested in learning more about what you are trying to do...

stuka · 04-24-2012, 05:09 AM

If you have a cluster you might want to look into the scatter/gather parallelism. Basically you do many GATK callings at once spread out around the cluster, with each job just doing a portion of the genomic intervals. Once all jobs are done you merge the results

http://www.broadinstitute.org/gsa/wi...er_parallelism

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
GATK variant calling on uniquely mapped reads?	pravee1216	Bioinformatics	5	03-26-2012 09:19 AM
GATK excludes some samples for cohort variant calling	liu_xt005	Bioinformatics	2	02-01-2012 12:58 PM
minimum depth variant calling samtools/gatk	m_elena_bioinfo	Bioinformatics	1	12-06-2011 09:31 AM
variant calling	kjaja	Bioinformatics	1	11-04-2011 08:16 AM
Variant Calling for Exome Capture Analysis	sbaheti	Bioinformatics	40	11-11-2010 11:35 AM

04-23-2012, 06:38 PM	#1
sehrrot Member Location: USA Join Date: Jul 2010 Posts: 58	GATK variant calling for multiple exome Hi all, Currently, I am running Unified genotyper for variant calling for around 100 exomes. It's on the cluster and I found the remaining running time is aournd 3 weeks. Is there any way that I can do within few hours? It's my first time to analyze high-throughput data.. Any advice would be appreciated

04-23-2012, 10:58 PM	#2
adaptivegenome Super Moderator Location: US Join Date: Nov 2009 Posts: 437	That seems like a long time. Can you share some more details about what you are doing so we can see if there is opportunity for speeding it up? Certainly a generic answer would be to leverage a cloud service to use as much CPU power as you need but I would be interested in learning more about what you are trying to do...

04-24-2012, 05:09 AM	#3
stuka Junior Member Location: OK Join Date: Oct 2008 Posts: 3	If you have a cluster you might want to look into the scatter/gather parallelism. Basically you do many GATK callings at once spread out around the cluster, with each job just doing a portion of the genomic intervals. Once all jobs are done you merge the results http://www.broadinstitute.org/gsa/wi...er_parallelism