bamqc steps request *a lot* of memory of open lsf

[malachig]

The following two steps (and possibly others) request 32Gb of memory:

BamQc - Collect Picard Metrics BamQc - Collect Picard G+C Bias

Since this is more than the VM has, these steps enter a pending state in the queue indefinitely. I can manually bmod the jobs but this raises a more general issue of perhaps having to do this all over the place. Any ideas on a more robust work around?

• Do we want to tell users that they will need at least 32Gb (or more) memory to do this test?
• Should we investigate whether BamQc really needs this much RAM?
• Is there a way to allow the user to set a $MAX_LSF_MEM environment variable that will globally reset resource requests that exceed this value. i.e. in workflow if a job is about to be submitted that asks for more than this, reset it to $MAX_LSF_MEM?

Other suggestions?

mg

[malachig]

This was fixed by reducing the memory requirement for this step by modifying:

Genome/Model/Tools/BamQc/Run.pm. It is used as max mem of lsf_resource for both picard CollectMultipleMetrics and CollectGcBiasMetrics.

Specifically by changing the default value for the parameter 'picard_maximum_memory'

Looks like it is set to request 16 Gb now and this seems to be working...