https://nf-co.re logo
Join Slack
Powered by
I'm working on getting GPU accelerated search work...
# proteinfold_dev
n
I'm working on getting GPU accelerated search working on Colabfold during this hackathon I'm running the pipeline from the current dev branch, but it seems to fail to pass the uniref database to the COLABFOLD_MMSEQS module. Interestingly, it seems like it's passing an empty channel to the module, rather than failing to find the files. It also looks like the prepare DB step is skipped entirely. Any ideas if this is a known issue?
Copy code
#!/bin/bash
export MODULEPATH="${MODULEPATH}:/home/z3545907/spack/share/spack/modules/"
module load linux-rocky8-x86_64_v2/nextflow/25.10.2 java/21.0.3_9-openjdk

SAMPLESHEET="$HOME/samplesheet.csv"
OUT_DIR="/srv/scratch/sbf/nathan/run_$PBS_JOBID/"

nextflow -c "./katana.config" run nf-core/proteinfold -r dev -latest \
    --input "${SAMPLESHEET}" \
    --outdir "${OUT_DIR}" \
    --mode "colabfold" \
    --use_gpu \
    --colabfold_db /srv/scratch/sbf-pipelines/proteinfold/dbs \
    --use_msa_server false \
    --colabfold_model_preset alphafold2_ptm \
    --colabfold_alphafold2_params_path /srv/scratch/sbf-pipelines/proteinfold/dbs/ \
    ${ARGS} \
    -profile apptainer
Copy code
File "/usr/local/lib/python3.12/dist-packages/colabfold/mmseqs/search.py", line 468, in main
      mmseqs_search_monomer(
    File "/usr/local/lib/python3.12/dist-packages/colabfold/mmseqs/search.py", line 93, in mmseqs_search_monomer
      raise FileNotFoundError(f"Database {db} does not exist")
  FileNotFoundError: Database uniref30_2302_db does not exist
Looking in the work directory, it looks like the database is not there:
Copy code
z3545907@k166:~/script $ ls -al /home/z3545907/script/work/ba/96866094693051ac6f530fdf4c26e9
total 66
drwxr-xr-x. 3 z3545907 unsw    12 Feb 26 09:44 .
drwxr-xr-x. 3 z3545907 unsw     3 Feb 26 09:44 ..
-rw-r--r--. 1 z3545907 unsw     0 Feb 26 09:44 .command.begin
-rw-r--r--. 1 z3545907 unsw   598 Feb 26 09:44 .command.err
-rw-r--r--. 1 z3545907 unsw   823 Feb 26 09:44 .command.log
-rw-r--r--. 1 z3545907 unsw   225 Feb 26 09:44 .command.out
-rw-r--r--. 1 z3545907 unsw 10645 Feb 26 09:44 .command.run
-rw-r--r--. 1 z3545907 unsw   376 Feb 26 09:44 .command.sh
-rw-r--r--. 1 z3545907 unsw     0 Feb 26 09:44 .command.trace
-rw-r--r--. 1 z3545907 unsw     1 Feb 26 09:44 .exitcode
lrwxrwxrwx. 1 z3545907 unsw    70 Feb 26 09:44 input.csv -> /home/z3545907/script/work/fb/e2445595a0de5ba5a627b6b2324e8b/input.csv
drwxr-xr-x. 2 z3545907 unsw    11 Feb 26 09:44 results
Copy code
[fb/e24455] process > NFCORE_PROTEINFOLD:COLABFOLD:MULTIFASTA_TO_CSV (MKEFWN)      [100%] 1 of 1 ✔
[ba/968660] process > NFCORE_PROTEINFOLD:COLABFOLD:MMSEQS_COLABFOLDSEARCH (MKEFWN) [  0%] 0 of 1 ✘
[-        ] process > NFCORE_PROTEINFOLD:COLABFOLD:COLABFOLD_BATCH                 -
[-        ] process > NFCORE_PROTEINFOLD:POST_PROCESSING:GENERATE_REPORT           -
[-        ] process > NFCORE_PROTEINFOLD:POST_PROCESSING:MULTIQC                   -
Execution cancelled -- Finishing pending tasks before exit
My bad - seems the issue was a symlink that was point to a non-existent directory. Please disregard 😅
j
No worries! That are the best issues to be reported, the already fixed ones 😄
😂 1
2 Views
Open in Slack
PreviousNext
Powered by