FASTQC fails with too little memory

[alexhbnr]

FASTQC fails with the following error message when being run after clipping:

Exception in thread "Thread-1" java.lang.OutOfMemoryError: GC overhead limit exceeded
        at java.lang.String.toCharArray(String.java:2899)
        at uk.ac.babraham.FastQC.Modules.PerBaseQualityScores.processSequence(PerBaseQualityScores.java:132)
        at uk.ac.babraham.FastQC.Analysis.AnalysisRunner.run(AnalysisRunner.java:89)
        at java.lang.Thread.run(Thread.java:748)

This might suggest that the amount of memory for this program should be increased to > 1GB.

Thanks,
Alex

[apeltzer]

Fixed in dev - can you check whether that helped?

[alexhbnr]

Thanks for the quick adaption. It now submits the job with more memory to SLURM, however, still fails with the same error message. If I ran it mainly and submit it with the same memory requirements, it magically works. According to the sacct output, it barely needed any memory, though:

   CPUTime     MaxRSS
---------- ----------
  00:02:31
  00:02:31    342844K

So I am wondering if I have some edge case files or something else breaks here.
Any ideas or suggestions? I am currently using FASTQC v0.11.8, in your docker file you state v0.11.9, but I doubt that it is the problem because for a subset of the samples it works totally fine.

[jfy133]

Could you point me privately which files you're trying, and I'll try it out myself to see if I see anything else.

I could also try changing the fastqc command to specifically define memory requirements with -Xmx

[jfy133]

@alexhbnr please try running your command with -r patch-fastqc. I now got your command to run successfully. I've made a PR into dev already, but it would be good for second replication.

[alexhbnr]

This patch fixes the issue. Thanks for catching this, @jfy133 !