Tutorial:Building OMA database on SGE

Question

[javier.herrero]

There is a buglet when running OMA on SGE. This is triggered when running it as a single job. For instance, one would want to do that for building the databases with "OMA -c" or for the last step after all the "OMA -s".

The error message is:

only_run_dbconv := false
Starting database conversion and checks...
ERROR: ParallelInfo expects a 1st argument, NrParallelProc:posint, found: undefined

The workaround is to run OMA as:

qsub -t 1 OMA

Javier

[GenoMax]

buglet - does that refer to a small bug in code :-)

Are you just reporting this as an observation or is there a question associated?

[javier.herrero]

Indeed, I am reporting this as an observation. I realise that this post is classified as a question, but I couldn't find any suitable category in the available options.

I do think this is a small bug as one might want to run OMA on an SGE cluster as:

'qsub OMA -c'

'qsub -t 1-10000 OMA -s'

'qsub OMA'

But this approach fails as when running on an SGE environment, OMA expects to find information about the number of parallel jobs (more precisely, the number of jobs in the job array). The workaround as I mentioned earlier is to run OMA as a job array of 1 single element. I thought it was useful to report it here in case someone is having trouble with this.

I hope this makes sense.

[GenoMax]

I have changed category to "Tutorial" (since that is the closest thing to describe this observation) and edited the title to provide more info. See if that works.

[javier.herrero]

If you think that is best, I am personally fine with that. That said, I am not sure I am the most suited person to provide a tutorial on running OMA. Maybe the OMA team might want to chip in.

[pourbiostars]

It works ! Thank you !