singularity pyscenic grn raises segmentation fault [BUG]

[biostu24]

Describe the bug When running pyscenic grn from a singularity container, a segmentation fault is raised. However, the output data is successfully generated.

This bug has been reported before but closed without addressing the bug, see #193. Although valid data is output from grn, this bug will cause pipelines using pyscenic to fail.

This bug is sporadic, for me all my runs have raised this error. For a colleague of mine, they have had some runs with and without the error.

Steps to reproduce the behavior

1. Command run when the error occurred: I am running pyscenic on a linux cluster using Slurm.

set -euo pipefail

singularity run aertslab-pyscenic-0.10.0.sife \
    pyscenic grn \
        --num_workers 16 \
        -o MyOutFile \
        MyInputMatrix \
        hs_hgnc_tfs.txt

3. Error encountered:

/opt/slurm/var/spool/slurmd/job1633626/slurm_script: line 32: 1603180 Segmentation fault      singularity run $singularity_image pyscenic grn --num_workers $n_workers -o $out_expr_adj_file $input_matrix "${scenic_data}hs_hgnc_tfs.txt"

Expected behavior No seg fault

Please complete the following information:

• pySCENIC version: 0.10.0
• Installation method: Singularity
• Run environment: Slurm (sbatch)
• OS: CentOS Linux 8
• Package versions: NA

[Jasoncbh]

I've had the same problem, and I don't know if the result is credible

[ghuls]

There is a new Docker image now with pySCENIC 0.12.0. The segfault is likely caused by the fact that you are running on an older CPU type which doesn't support certain instructions.

[biostu24]

There is a new Docker image now with pySCENIC 0.12.0. The segfault is likely caused by the fact that you are running on an older CPU type which doesn't support certain instructions.

Let’s hope this new image resolves the issue. We relatively new hardware (< 3years) so I doubt that is the problem.