How-To: Running Alphafold3 on Talapas

An Alphafold3 module is available on Talapas. For convience and ease of use we have included some scripts to make running the Alphafold3 pipeline straightforward, but the containerized alphafold3 can also be run directly.

An Alphafold3 image file can be found at:

/packages/alphafold3/3.0.0/alphafold3.sif

This image file can be run with Apptainer/Singulatrity. For example,

apptainer exec \
     --nv \
     --bind $HOME/af_input:/<UID>/af_input \
     --bind $HOME/af_output:/<UID>/af_output \
     --bind /packages/alphafold3/models:/home/<UID>/models \
     --bind /packages/alphafold3/public_databases:/home/<UID>/public_databases \
     alphafold3.sif \
     python /packages/alphafold3/3.0.0/src/run_alphafold.py \
     --json_path=<INPUT_FILE_LOCATION> \
     --model_dir=/packages/alphafold3/models \
     --db_dir=/packages/alphafold3/public_databases \
     --output_dir=<OUTPUT_LOCATION>

When you run module load alphafold3/3.0.0 in the command line, some helper scripts will be added to your path. Using the alphafold3 helper method will run the container and set the correct paths so Alphafold3 can use the public databases and model paramters centrally stored on Talapas. You will still have to supply inputs for the path to the input directory (or input json) and the desired output directory path:

alphafold3 \
    --json_path=/path/to/input_file.json \
    --output_dir=/home/<UID>/af_out/ [other options]

Talapas Knowledge Base

How-To: Running Alphafold3 on Talapas

Analytics

Related articles