MISO on a SLURM based cluster?

[kben]

In order to run MISO (http://genes.mit.edu/burgelab/miso/docs/) on a SLURM (https://computing.llnl.gov/linux/slurm/) operated cluster any tips facilitating multiple CPU usage would be awesome! At the moment I can't motivate MISO to use all of the "booked" node CPUs.

job script:

#!/bin/bash
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=24
#SBATCH --error=miso_err.log
#SBATCH --output=miso_out.log
#SBATCH --job-name=miso_laczA
#SBATCH --mem-per-cpu=3000
#SBATCH --partition=test
#SBATCH --time=00:59:59
#
export OMP_NUM_THREADS=24
module load python/2.7.3
python -c "import numpy"
python -c "import scipy"
python -c "import pysam"
srun python $HOME/programs/MISO/misopy/run_events_analysis.py --compute-genes-psi /MISO/hg19/indexed/indexedSE /bam/laczA/Aligned.out.sorted.bam --output-dir laczA_MISO_output/ --read-len 100 --paired-end 186 16 --use-cluster --chunk-jobs 200

Available MISO options are:

--use-cluster
--chunk-jobs N
and some for SGEarray
(http://genes.mit.edu/burgelab/miso/d...s-to-a-cluster)

Thanks for any idea!

Ben

[kben]

Maybe MISO only uses full nodes? So

#SBATCH --nodes=x

might work? I'll try.

[kben]

Now testing "--chunk-jobs N" as SLURM tasks (instead of SLURM threads).

[soren.lykkeandersen]

Did you find a solution?