Apptainer wrapper commands for any system

Running without using pre-installed HPC modules

If you are running the command from this tutorial on a computer other than the Monash M3 HPC cluster, you won’t necessarily have the same pre-installed modules (eg module load).

Here are more detailed snippets that contain the additional Apptainer prefixes required for each command to use the containers available via our container registry on Github.

These should run anywhere with Apptainer installed (and a supported NVIDIA GPU).

You’ll also need Nextflow for the nf-binder-design examples - see Appendix: Foundational skills: Nextflow crash course.

RFdiffusion + ProteinMPNN + AF2 initial guess

# This prefix `PREFIX_RFD` is here to run RFDiffusion via an Apptainer container. 
# It should work anywhere where Apptainer is installed. You'll need an NVIDIA GPU. 

# This version pulls the container image from the Github Package Registry - it should work on any system with apptainer installed. The first time it's run there's an initial delay while the container image is downloaded.
PREFIX_RFD="apptainer exec --nv -B $(mktemp -d):/usr/local/lib/python3.10/dist-packages/schedules docker://ghcr.io/australian-protein-design-initiative/containers/rfdiffusion:pytorch2407 "

# This version uses an Apptainer image pre-downloaded to the workshop scratch space
#PREFIX_RFD="apptainer exec --nv -B $(mktemp -d):/usr/local/lib/python3.10/dist-packages/schedules /scratch2/pd27/shared/containers/ghcr.io-australian-protein-design-initiative-containers-rfdiffusion-pytorch2407.img "

mkdir -p output/rfdiffusion

$PREFIX_RFD /app/RFdiffusion/scripts/run_inference.py \
  inference.input_pdb=input/PDL1.pdb \
  'contigmap.contigs=[A18-132/0 65-120]' \
  'ppi.hotspot_res=[A56]' \
  inference.output_prefix=output/rfdiffusion/pdl1_test \
  inference.num_designs=4 \
  denoiser.noise_scale_ca=0 \
  denoiser.noise_scale_frame=0

# This version pulls the container image from the Github Package Registry
PREFIX_PMPNN="apptainer exec --nv docker://ghcr.io/australian-protein-design-initiative/containers/proteinmpnn_dl_binder_design:latest "

# This version uses an Apptainer image pre-downloaded to the workshop scratch space
#PREFIX_PMPNN="apptainer exec --nv /scratch2/pd27/shared/containers/ghcr.io-australian-protein-design-initiative-containers-proteinmpnn_dl_binder_design-latest.img "

mkdir -p output/proteinmpnn

$PREFIX_PMPNN /app/dl_binder_design/mpnn_fr/dl_interface_design.py \
    -pdbdir input/ \
    -relax_cycles 0 \
    -seqs_per_struct 2 \
    -outpdbdir output/proteinmpnn/ \
    -omit_AAs C

PREFIX_AF2IG="apptainer exec --nv docker://ghcr.io/australian-protein-design-initiative/containers/af2_initial_guess:nv-cuda12 "

mkdir -p output/af2_initial_guess/pdbs

$PREFIX_AF2IG python /app/dl_binder_design/af2_initial_guess/predict.py \
    -pdbdir output/proteinmpnn \
    -outpdbdir output/af2_initial_guess/pdbs/ \
    -recycle 3 \
    -scorefilename output/af2_initial_guess/pdl1_test.scores.cs

BindCraft

export APPTAINERENV_PREPEND_PATH=/opt/conda/envs/BindCraft/bin/

# This version pulls the container image from the Github Package Registry
PREFIX_BINDCRAFT="apptainer exec --cleanenv --env MPLCONFIGDIR=/.matplotlib -B $(mktemp -d):/.matplotlib --nv docker://ghcr.io/australian-protein-design-initiative/containers/bindcraft:05702c4_nv-cuda12 "

# This version uses an Apptainer image pre-downloaded to the workshop scratch space
#PREFIX_BINDCRAFT="apptainer exec --cleanenv --env MPLCONFIGDIR=/.matplotlib -B $(mktemp -d):/.matplotlib --nv /scratch2/pd27/shared/containers/ghcr.io-australian-protein-design-initiative-containers-bindcraft-05702c4_nv-cuda12.img "

#    # Ensure that BindCraft finds the correct version of ffmpeg
#    export PATH=/opt/conda/envs/BindCraft/bin/:\$PATH

$PREFIX_BINDCRAFT /opt/conda/envs/BindCraft/bin/python \
    /app/BindCraft/bindcraft.py \
        --settings settings.json \
        --filters /app/BindCraft/settings_filters/default_filters.json \
        --advanced /app/BindCraft/settings_advanced/default_4stage_multimer.json

# 
#         --filters /app/BindCraft/settings_filters/default_filters.json \

Reuse

CC BY 4.0