This repository serves to reference the data and code of co-folding models used in the study "Do Deep Learning Models for Co-Folding Learn the Physics of Protein-Ligand Interactions?"

AlphaFold3 is available via the repository here: https://github.com/google-deepmind/alphafold3. Version 3.0.0 was used in the study. Following their installation instructions, you can run the test example by running the following Docker command:

docker run -it \
    --volume examples/af_input:/root/af_input \
    --volume examples/af_output:/root/af_output \
    --volume <MODEL_PARAMETERS_DIR>:/root/models \
    --volume <DATABASES_DIR>:/root/public_databases \
    --gpus all \
    alphafold3 \
    python run_alphafold.py \
    --json_path=/root/af_input/fold_input.json \
    --model_dir=/root/models \
    --output_dir=/root/af_output

RFAA is available via the repository here: https://github.com/baker-laboratory/RoseTTAFold-All-Atom. Version 6c85140 was used in the study. Following their installation instructions, you can run the test example by running the following commands:

cp examples/rfaa_input/example.yaml YOUR_CONFIG_PATH/.
python -m rf2aa.run_inference --config-name example

Chai-1 is available via the repository here: https://github.com/chaidiscovery/chai-lab. Version 0.5.0 was used in the study. Following their installation instructions, you can run the test example by running the following command:

chai-lab fold --use-msa-server --use-templates-server examples/chai_input/example.fasta output_folder

The latest version of Boltz-1 is available via the repository here: https://github.com/jwohlwend/boltz. Version 0.4.1 was used in the study. Following their installation instructions, you can run the test example by running the following command:

boltz predict examples/boltz_input/example.yaml --use_msa_server

All models were run on NVIDIA GPU hardware with at least 12GB of RAM.