Visu to helix

[ChristineSchulz]

• Tutorial for transitioning from bwVisu to bwForCluster Helix using the Boltz2 example

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[iulusoy]

So I got stuck after running the first colabfold job on helix. It only created one output file (*.a3m). I assume that I need to rerun the Boltz tutorial on bwVisu to get the correct files (I still have insulin.fasta but I think it was renamed here, or there are some unmerged changes from main (http://127.0.0.1:8000/tutorials/tutorial_Boltz_bwVisu/#step-4-set-a-working-directory-and-upload-files) also references a notebook that does not exist anymore).

Is it possible to show all file content here again/link it, so I can copy it again or verify everything is correct?

[iulusoy]

run_msa.slurm:

#!/bin/bash
#SBATCH --partition=gpu-single
#SBATCH --ntasks=1
#SBATCH --time=00:20:00
#SBATCH --gres=gpu:A100:1,gpumem_per_gpu=40GB
#SBATCH --mem=8gb

ws_allocate your_work_space 30
RESULTS_DIR=`ws_find your_work_space`

module load devel/miniforge/24.9.2
module load devel/cuda/12.8
conda activate /mnt/sds-hd/sd25g005/boltz

export PATH="/mnt/sds-hd/sd25g005/boltz/localcolabfold/colabfold-conda/bin:$PATH"

colabfold_search  \
    --db-load-mode 2 \
    --threads 96 \
    --use-env 0 \
    --gpu 1 \
    "/home/hd/hd_hd/hd_lt239/biostructurehub/tutorials/new_boltz/insulin.fasta" \
    "/mnt/sds-hd/sd25g005/boltz/localcolabfold" \
    $RESULTS_DIR \

[iulusoy]

insulin.fasta:

>sp|P01308|INS_HUMAN Insulin OS=Homo sapiens OX=9606 GN=INS PE=1 SV=1
MALWMRLLPLLALLALWGPDPAAAFVNQHLCGSHLVEALYLVCGERGFFYTPKTRREAED
LQVGQVELGGGPGAGSLQPLALEGSLQKRGIVEQCCTSICSLYQLENYCN

[iulusoy]

Output is sp_P01308_INS_HUMAN_Insulin_OS_Homo_sapiens_OX_9606_GN_INS_PE_1_SV_1.a3m and no other files, while the slurm output shows no errors or warnings.

[ChristineSchulz]

That sounds / looks good so far!

The colabfold runs mmseqs2 which does the multisequence alignment, resulting in the .a3m file (see image here http://127.0.0.1:8000/tutorials/tutorial_Boltz_bwVisu/#verify-input). Insulin.fasta should remain the same as before.

[ChristineSchulz]

The .yaml for Insulin should be

version: 1
sequences:
  - protein:
      id: [A] 
      sequence: MALWMRLLPLLALLALWGPDPAAAFVNQHLCGSHLVEALYLVCGERGFFYTPKTRREAEDLQVGQVELGGGPGAGSLQPLALEGSLQKRGIVEQCCTSICSLYQLENYCN
      msa: $RESULTS_DIR/sp_P01308_INS_HUMAN_Insulin_OS_Homo_sapiens_OX_9606_GN_INS_PE_1_SV_1.a3m

[ChristineSchulz]

references a notebook that does not exist anymore

In http://127.0.0.1:8000/tutorials/tutorial_Boltz_bwVisu/#step-3-set-a-working-directory-and-upload-files I mention Boltz_w_mmseqs.ipynb which should be here https://github.com/ssciwr/BioStructureHub/blob/main/notebooks/Boltz_w_mmseqs.ipynb

Below I mention Boltz_Confidence_Levels.ipynb, which is here https://github.com/ssciwr/BioStructureHub/blob/main/notebooks/boltz_confidence_levels.ipynb

I am not sure which notebook you are referring to.

[ChristineSchulz]

Is it possible to show all file content here again/link it, so I can copy it again or verify everything is correct?

I have examples for the files in https://github.com/ssciwr/BioStructureHub/tree/main/references/boltz .