LearnReadOrientationModel is a Memory Hog

[DarioS]

I am doing somatic variant calling for whole genome sequencing. This tool could be improved to have reasonable RAM demands.

#PBS -l walltime=05:00:00,ncpus=26,mem=750GB
#PBS -q hugemem

Fails after almost two hours.

   NCPUs Requested:    26                     NCPUs Used: 26
                                           CPU Time Used: 28:09:38
   Memory Requested:   750.0GB               Memory Used: 749.95GB
   Walltime requested: 05:00:00            Walltime Used: 01:54:14

The reason is that the RAM allowance was exceeded.

Job 51320644 has exceeded memory allocation on node gadi-hmem-clx-0005.gadi.nci.org.au
Process "bash", pid 742028, rss 3346432, vmem 21680128
Process "51320644.gadi-p", pid 742086, rss 3059712, vmem 10174464
Process "mpirun", pid 742112, rss 6406144, vmem 220180480
Process "nci-parallel", pid 742118, rss 266031104, vmem 1423237120
    ...        ...

Normals about 30 times coverage, tumours about 60 times coverage, 26 patients in total. It should be made more efficient.