[miso-users] run-events-analysis stuck at "Waiting on 8 threads..."

[Yarden Katz]

Hi Maayan,

Could you please include the output?  The command you showed does not use "--use-cluster", so all it would do is use 8 virtual threads (since you used "-p 8") to split the jobs.  I'm not sure why it got stuck -- what does stdout/stderr say for individual genes/events?  Which MISO version are you using?  Having the output would help debug, or a small BAM file example to try out that reproduces this behavior.  

The "-p" option and "--use-cluster" are mutually exclusive, so to run on cluster you just drop "-p" and use "--use-cluster" (with "--chunk-jobs", optionally.)  You have to make sure your cluster submission command is set correctly in the settings file.  The default is "qsub" or "bsub" so all MISO would do is split the genes/events into chunks (specified by --chunk-jobs) and make those into bash scripts, one bash script for each job.  Then it would use the command in the settings file to launch these jobs, e.g. "bsub batch_1.sh".  So the command in the settings file needs to be such that it takes as argument a bash script and submits it to the cluster, using whatever procedure your system uses, since MISO cannot support the nuances of all cluster submission systems which vary widely across servers.

Best, --Yarden

On Oct 11, 2013, at 5:55 PM, Maayan Kreitzman wrote:

> Hi Yarden and all,
> I've previously run MISO successfully, (and it was really fast) but some 
> things seem to have changed with the multithreading and cluster submission.
> 
> For running on a single host, it gets stuck at the "Waiting on 8 
> threads..." step, and it's been there for about a week.
> 
> run_events_analysis.py --compute-genes-psi /testing/MISO/ENS_new/ accepted_hits.bam --output-dir $outdir --read-len 42 –prefilter -p 8
> 
> 
> For running on a cluster, are you supposed to run the same command 
> (without the -p 8 and with the --use-cluster) on the head-node? 
> Previously I just submitted the job to the cluster myself, but the docs 
> imply that miso will do the submitting.
> 
> thanks,
> maayan
> 
> -- 
> Maayan Kreitzman
> Computional Biologist
> Michael Smith Genome Sciences Centre
> BC Cancer Agency
> Vancouver, BC V5Z 4S6
> 
> 
> 
> _______________________________________________
> miso-users mailing list
> miso-users at mit.edu
> http://mailman.mit.edu/mailman/listinfo/miso-users