RE: IsONclust on ONY; spot died with <Signals.SIGILL: 4>.

[Malabady]

Hi I am running isONclust in a Conda environment and the spoa and medaka modules are loaded before the run.

Here is my commands:

ml Miniconda3/4.10.3
source activate isonclust
ml spoa/4.0.0-GCC-8.3.0
ml minimap2/2.22-GCC-8.3.0
ml medaka/1.5.0

isONclust --ont --use_old_sorted_file --fastq ./isONclust-out2/sorted.fastq  --outfolder isONclust-out2 --consensus  --t 24

the run failed in the consesus step when running spoa with the following kill signal.

Traceback (most recent call last):
  File "/home/malabady/.conda/envs/isonclust/bin/isONclust", line 263, in <module>
    main(args)
  File "/home/malabady/.conda/envs/isonclust/bin/isONclust", line 127, in main
    center = consensus.run_spoa(reads_path.name, os.path.join(work_dir,"spoa_tmp.fa"), "spoa")
  File "/home/malabady/.conda/envs/isonclust/lib/python3.10/site-packages/modules/consensus.py", line 89, in run_spoa
    subprocess.check_call([ spoa_path, reads, "-l", "0", "-r", "0", "-g", "-2"], stdout=output_file, stderr=null)
  File "/home/malabady/.conda/envs/isonclust/lib/python3.10/subprocess.py", line 369, in check_call
    raise CalledProcessError(retcode, cmd)
subprocess.CalledProcessError: Command '['spoa', '/tmp/tmp_12sh3ko/reads_tmp.fq', '-l', '0', '-r', '0', '-g', '-2']' died with <Signals.SIGILL: 4>.

I can see /tmp/tmp_12sh3ko/reads_tmp.fq and can run spoa on the reads_tmp.fq directly.

Any suggestions what's wrong?

Thanks Magdy

[ksahlin]

Hi @Malabady,

The --consensus parameter was an experimental parameter added after publishing isONclust. It has not been tested well and I would simply suggest not to use it in isONclust.

If you are looking for clustering + consensus forming + barcode removal and some more, you may try NGSpeciesID which is based on isONclust clustering method but adds some features on top of the clustering.

Finally, you should probably not call the --consensus if you are clustering transcript data because the consensus step will not take into account exon differences between transcripts. It is possible you can achieve some level of transcript level consensus forming with NGSpeciesID by setting --aligned_threshold and --mapped_threshold very high (like 0.95) but this is not the intended use of NGSpeciesID and can produce a fairly suboptimal result (depending on the nature of your data).

Best. K

[ksahlin]

...and as for the actual runtime error I don't know. As you mention, if you are able to run spoa without error on the instance reported in the error message ('spoa', '/tmp/tmp_12sh3ko/reads_tmp.fq', '-l', '0', '-r', '0', '-g', '-2') it is strange.

I would have expected that this instance would be the command throwing the error.