SGE scripts

[davidvilanova]

Hi, Does anybody uses SGE queing system ??? Actually the code runs on LSF. Would be great to have SGE calls also ?

[nmb85]

I adapted the create_jobs.py and setupDirs.py scripts for SGE on our cluster. This won't work automatically for you, but it should be much closer to working order on SGE than the original LSF scripts. Just replace the original copies of these scripts in the LSFScripts/ directory with these versions. I've replaced all the options and it works on my cluster, but you'll have to at least change the queue name ('micro') with your own queue name. Not sure if there are other changes you need to make...

https://github.com/russianconcussion/lsa.sge.scripts.git

[davidvilanova]

Thanks !!! that´s great. To quickly replace the (´micro´) queuing name i just write a sed line as follows sed "s/micro/myqueue/g" create_jobs.py > out1 sed "s/micro/myqueue/g" setupDirs.py > out2

thanks !!!

[nmb85]

Glad to help. Let me know if it doesn't work.

[davidvilanova]

By the way, how do you handle the HashCounting.sh parallel_gnu with SGE ??? The only way i food was to write a qsub script and call the HashCounting.sh script from inside. I don´t know how much memory need to be allocated, depending on the hashsize:

scriptSGE.sh

$ -cwd

$ -l mem=28G

$ -o out

$ -e err

$ -q queuelocal

hashSize=22 python LSA/create_hash.py -i original_reads/ -o hashed_reads/ -k 33 -s $hashSize > Logs/CreateHash.log

did you find a better way ??

[nmb85]

I'm not sure that I even ran that script. I followed the large dataset analysis instructions at http://latentstrainanalysis.readthedocs.org/en/latest/large_collections.html. The instructions never said to run that script, and I don't have any record of running that script in my logs. Sorry, but I can't help there!