Broad best practices workflow in Galaxy

[KevinLam]

I read about some users trying to recreate the Broad's workflow for SNPs within a local galaxy install. Did anyone have success with it?
How might I do this with the least amount of effort?

are there virtualbox images or AMI for such a server?

[dawe]

Hi there, we are working on it also. The main problem we have is that we cannot achieve a reasonable level of parallelism (i.e. we cannot split GATK bam Walkers to run by chromosome)

[nilshomer]

Hi there, we are working on it also. The main problem we have is that we cannot achieve a reasonable level of parallelism (i.e. we cannot split GATK bam Walkers to run by chromosome)

For the unified genotyper: "The -L option lets you specify the region to process". So it would seem that there is some functionality for splitting up the data by region.

[RDW]

You probably want to use GATK's scatter/gather parallelism:

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

Their 'best practice' Queue script is set up to implement this, but you'll need a job runner like GridEngine installed to make use of it:

http://www.broadinstitute.org/gsa/wi...ssing_Pipeline
http://www.broadinstitute.org/gsa/wi...th_Grid_Engine

However, this may be a pain to set up:

http://seqanswers.com/forums/showthread.php?t=18075

and the canned script only takes you as far as analysis-ready cleaned BAM files, not called variants.