Protenix: Protein + X Dataset Overview

Dataset Content

• Data Source: wwPDB dataset
• Data Type: Protein structure data
• Data Size: Requires at least 1 T of disk space
• Data Structure:

  ├── components.v20240608.cif [408M] # ccd source file
  ├── components.v20240608.cif.rdkit_mol.pkl [121M] # ccd source file generated by rdkit Mol object
  ├── indices [33M] # chain or interface entries
  ├── mmcif [283G]  # original mmcif data
  ├── mmcif_bioassembly [36G] # pre‑processed wwPDB structural data
  ├── mmcif_msa [450G] # msa files
  ├── posebusters_bioassembly [42M] # pre‑processed posebusters structural data
  ├── posebusters_mmcif [361M] # original mmcif data
  ├── recentPDB_bioassembly [1.5G] # pre‑processed recentPDB structural data
  └── seq_to_pdb_index.json [45M] # sequence‑to‑pdb‑id mapping file
  

Data Download

• Pre‑processed data download:

  wget -P /af3-dev/release_data/ https://af3-dev.tos-cn-beijing.volces.com/release_data.tar.gz
  tar -xzvf /af3-dev/release_data/release_data.tar.gz -C /af3-dev/release_data/
  rm /af3-dev/release_data/release_data.tar.gz
  

• Inference‑only data download:

  wget -P /af3-dev/release_data/ https://af3-dev.tos-cn-beijing.volces.com/release_data/components.v20240608.cif
  wget -P /af3-dev/release_data/ https://af3-dev.tos-cn-beijing.volces.com/release_data/components.v20240608.cif.rdkit_mol.pkl
  

Model Checkpoints

• Pre‑trained model download:

  wget -P /af3-dev/release_model/ https://af3-dev.tos-cn-beijing.volces.com/release_model/model_v1.pt
  

Dataset Uses

• Training: From‑scratch model training
• Inference: Model inference and prediction

Dataset Processing

• Data processing scripts: Being organized and prepared; distilled data will be released in the future.

Related Documentation

• Input JSON file format: Details
• Training and fine‑tuning settings: Details

License

• Non‑commercial use: Creative Commons Attribution‑NonCommercial 4.0 International License
• Commercial use: Contact ai4s‑bio@bytedance.com for a commercial license