BiocParallel Errors on HPC

[OdmaaBayar]

The bambu step fails consistently during read class file generation with BiocParallel errors inside the Singularity container (Stop worker failed with the error: wrong args for environment subassignment). I am running GoekeLab/bambu-singlecell-spatial (rev 38eb5de) on Nextflow 25.10.0 with 16–32 cores, demultiplexed BAMs (barcodes present, no UMIs, --deduplicateUMIs FALSE). The error appears at ~5–11% progress before extended annotation, suggesting a BiocParallel Multicore/fork backend crash. What is the best approach to resolve this? Are there any other additional parameters I need to modify to get this to run on HPC?

Thank you!

[lingminhao]

Hi @OdmaaBayar ,

May I have the code you used to run, we was trying to check whether this is an issue when running bambuSC on multiple samples.

Best Regards,
Min Hao

[OdmaaBayar]

Hi @lingminhao,

Here is my shell script that I used to run Bambu:

#!/bin/bash
#SBATCH --job-name=bambu_sc
#SBATCH --nodes=1
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=16
#SBATCH --mem=300G
#SBATCH --time=10-00:00:00

# Example Nextflow command for running Bambu single-cell/spatial workflow
nextflow run GoekeLab/bambu-singlecell-spatial \
  --bams samplesheet_bam.csv \
  --deduplicateUMIs FALSE \
  --genome /path/to/reference/genome.fa \
  --annotation /path/to/reference/annotation.gtf \
  --ncore 16 \
  --outdir bambu_output \
  -with-singularity 'docker://lingminhao/bambusc:beta1.2'

These mapped bam files were generated using human genome (Hg38) and annotation (v39) so I'm also wondering if my choice of Gencode's genome.fa (v39) and annotation.gtf (v39) is contributing to the problem. Please let me know if you have any suggestions.

Thank you!