Pulling Container with Apptainer at BASECALL:DORADO_BASECALL:downloadModel step fails #119

ashdederich · 2025-05-30T23:26:17Z

I am working to get the Master of Pores pipeline running at my HPC. Apptainer is our HPC's preferred method for containerization. I have tried running the MOP pipeline as such:

module load nextflow
module load apptainer
module load cuda/11.8.0

nextflow run mop_preprocess.nf -with-apptainer -params-file params.yaml -w $scratchdir -profile slurm

Pulling the container at the BASECALL:DORADO_BASECALL:downloadModel step always fails, with this error message:

Command error:
  INFO:    gocryptfs not found, will not be able to use gocryptfs
  FATAL:   container creation failed: while getting /proc/self/ns/user file descriptor: while requesting file descriptors send: read unix @->@: read: connection reset by peer

However, if I run the command like this:

module load nextflow
module load singularity
module load cuda/11.8.0

nextflow run mop_preprocess.nf -with-singularity -params-file params.yaml -w $scratchdir -profile slurm

The container successfully downloads. My question is - Nextflow supports of apptainer, and reading through the MOP docs, it looks like MOP supports the use of apptainer. So why does the container creation step fail 100% of the time when I change to -with-apptainer?

Lastly, I'll add that, with the -with-singularity flag, the MOP pipeline fails a few steps further (due to a versioning issue with singularity). Due to this error and my HPC's strong preference for apptainer overs singularity, I switched to trying -with-apptainer, instead.

Here is my params.yml file:

pod5: "/scratch/general/nfs1/fleming_mop_input/pod5_converted_files/*.pod5"
fastq: ""
## This can be empty but then you need to add specify kit and flowcell via command line inside pars_tools
conffile: "final_hek_tnfa_1-1.txt"
# Basecalling can be either NO, dorado, dorado-mod or dorado-duplex
basecalling: "dorado"
## Can be OFF / cuda10 / cuda11. Newer version of GUPPY may require cuda11
GPU: "cuda11"
# OFF may not be legitimate options for demultiplexing and demulti_fast5.
demultiplexing: "NO"
demulti_fast5: "OFF"
demulti_pod5: "OFF"
### Number of fast5 basecalled per parallel job
granularity: 1
### File with the list of accepted barcodes. It can be empty
barcodes: ""

# Needed for fastq input
## fast5: ""

# Common
reference: "/scratch/general/nfs1/u0432542/all_ref_rrna_human.fasta"
## Can be transcriptome / genome
ref_type: "transcriptome"
annotation: ""
## command line options
pars_tools: "${projectDir}/tool_opts/drna_tool_splice_opt.tsv"
## Cut off quality for QC
qualityqc: 5
## Can be nanoq / nanofilt
filtering: "nanoq"
## Can be graphmap / graphmap2 / minimap2 / bwa
mapping: "minimap2"
## Can be nanocount for transcriptome / htseq for genome
counting: "nanocount"
## Can be NO / bambu / isoquant
discovery: "NO"
## Convert bam to cram
cram_conv: "NO"
subsampling_cram: 50

progPars:
  basecalling:
    dorado: "fast"
    dorado-duplex: ""
    dorado-mod: ""
  demultiplexing:
    seqtagger: ""
    dorado: ""
  filtering:
    seqkit: ""
    nanoq: ""
  mapping:
    graphmap: ""
    graphmap2: ""
    minimap2: ""
    bwa: ""
    winnowmap: ""
  counting:
    htseq: ""
    nanocount: ""
  discovery:
    bambu: ""

The text was updated successfully, but these errors were encountered:

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Pulling Container with Apptainer at BASECALL:DORADO_BASECALL:downloadModel step fails #119

Pulling Container with Apptainer at BASECALL:DORADO_BASECALL:downloadModel step fails #119

Pulling Container with Apptainer at BASECALL:DORADO_BASECALL:downloadModel step fails #119

Pulling Container with Apptainer at BASECALL:DORADO_BASECALL:downloadModel step fails #119

Comments