icebert / pblat-cluster Goto Github PK

Hello Meng,

Great tool!! The cluster version is quick and accurate.
I have no issues using the cluster version against the genome file size of around 6MB and reads file size of 800MB with 149 read sequences. It does the analysis in about 3-4 hours.
However, I now have a genome file of size around 30MB and reads file of size around 1GB (with 49 read sequences). I left it to run for a day or so and it's still running. Sometimes it just crashes. The files in question can be found in the links below:
https://www.mediafire.com/file/v5sgvajfnweeih7/genome.fa/file
https://www.mediafire.com/file/64jiwm4cgn0bbuq/reads.fa/file

Could you please advice as to what would be the ideal ratio of nodes/cores and memory to get this analyzed as fast as possible. It still doesn't finish even if I increase the cores/nodes and memory.
I use the following sbatch script:

#SBATCH -N 1
#SBATCH -n 8
#SBATCH --mem=14G
#SBATCH --time=06:00:0
module load nixpkgs/16.09  gcc/7.3.0  openmpi/3.1.4
mpirun ./pblat-cluster genome.fa reads.fa -out=pslx out.pslx

The above script works well on the 6MB genome and 800MB reads files.

Thanks,
Vijay