agat_sp_split_by_level2_feature Bus error

[ViriatoII]

Hi!

I keep having this Bus error. Is it related to memory? I'm using 20 Gbs memory in a PBSpro cluster.

This happened sometimes when using agat installed with conda and also when running your nextflow pipeline.

agat_sp_split_by_level2_feature.pl -g MPI_run.all.gff -o maker_results_noAbinitio_clean
Command error:
  .command.sh: line 2: 86004 Bus error   

error exit status (135)

[ViriatoII]

I think it has to do with the output directory already existing (often when the command ran unsuccessfuly before and created an empty folder).

The function doesn't know how to deal with this and crashes in a bus error.

[Juke34]

I don't think the problem is related to AGAT. You can try to run the command standalone (outside the nextflow pipeline) to check if it is really related to AGAT. (you must activate the AGAT Conda env which is somewhere is the work dir, or install your own AGAT conda env)

[ViriatoII]

Hi!
Standalone now it worked. Generated 2 files:
1181933698 22. Apr 13:34 match_part.gff
72231711 22. Apr 13:34 mrna.gff

So is this a nextflow problem? That pipeline comes with it's own version of agat and automatically activates that environment, as I can see:

source $(conda info --json | awk '/conda_prefix/ { gsub(/"|,/, "", $2); print $2 }')/bin/activate home/annotation_project/scratch/nxf_work/conda/agat-6b6131ed4ecf41887a64062ec60a9397  

But I remember having this bus error a couple of months ago, and then I was only using agat standalone.

[Juke34]

Ok probably memory issue. All AGAT scripts with _sp_ in the name load the file into memory, that could end to take ~6 time the size of the file in memory. So for 1Gb file it can take ~6 Gb in memory. It sounds you set up 2gb in your pbs profile, and your file is big. So there is not enough memory apparently. Either increase the memory size, or clean your GFF file first. Indeed it sounds to contain lot of unnecessary data (what end up in the match file). To clean it use a awk command and keep only what is maker in the 3rd column.

[Juke34]

Sorry I meant 2nd column!

[Juke34]

I moved your last comment in a new issue (#37).
I consider this one resolved.