Beta SLURM support in MaSuRCA 3.3.0b

I have been working to implement grid execution in MaSuRCA for a while, and MaSuRCA supported SGE for nearly a year now.  The implementation of support for SLURM was behind due to an important difference between job submission in SGE and SLURM: absence of "sync" option, where the submit command in a script would not return until all jobs have exited.  There are ways around it and we are still testing what would work best in the framework of MaSuRCA.  For now one can run the four most computation intensive parts of MaSuRCA on the grid:

1. mega-reads correction 1st pass
2. mega-reads correction second pass (sometimes grid is not necessary for this one)
3. overlap-based trimming in CABOG
4. overlapping in CABOG

The way this works, one runs assemble.sh and then the run prepares batch jobs and exits, instructing user to run a given sbatch command with given parameters.  Once all sbatch jobs are done, one should re-run assemble.sh script (would not hurt to re-generate it just in case).  The scrip checks if all jobs completed successfully and then proceeds.  If some jobs have failed, the script exits, instructing the user to run sbatch command again.

To configure the grid run one must add the following options to the config file.  Hint: these options are already present in the example config file created upon installation of MaSuRCA

USE_GRID=1
GRID_QUEUE=your_slurm_partition
GRID_ENGINE=SLURM
GRID_BATCH_SIZE=<amount of Nanopore/Pacbio sequence to use per batch>, 300000000 default, will create 10 batches for every 1x of mammalian genome coverage
NUM_THREADS=32

The sbatch command will run one job per 32-core node in this case.  If your nodes have fewer or more cores, adjust accordingly.

If grid jobs fail in mega-reads pass2 (jf_aligner jobs) it is likely due to not enough memory on the nodes.  To solve this, reduce GRID_BATCH_SIZE, re-generate assemble.sh , remove mr_pass2 folder and re-run.

To get the 3.3.0b version, please go to github
https://github.com/alekseyzimin/masurca/blob/master/MaSuRCA-3.3.0b.tar.gz

Thanks to Daniela Puiu, the software engineer at the Center for Computational Biology for implementing SLURM support in CABOG and helping me test MaSuRCA!