Chai
1. Introduction
Chai-11 is a multimodal molecular structure prediction foundation model, focusing on accurate 3D structure prediction of biomolecules. It achieves state-of-the-art performance in drug discovery and biomolecular interaction studies. Its core value lies in using deep learning techniques to decode the folded structures and interaction mechanisms of biomolecules, providing critical support for targeted drug design and protein function studies.
Chai-1 efficiently handles a variety of biomolecular structure prediction tasks, including:
- Protein–ligand (small molecule) complex structure prediction;
- Protein multimer (e.g., antibody–antigen, protein–protein complex) structure prediction;
- Protein monomer, nucleic acid (DNA/RNA) structure prediction, and more.
Figure 1. Overall architecture of Chai-1
Figure 2. Performance of Chai-1
2. Parameter Description
This tool supports multiple molecule types, including proteins, DNA, RNA, and small molecules. You can select the type via the Molecule Type parameter. Currently, only standard amino acids, standard bases, and small molecules in SMILES format are supported.
Click the Add Sequence button to add a new sequence. Chains will be automatically labeled using uppercase letters A–Z based on the order of input.
Click the Add Constraints button to incorporate different types of constraints, such as:
- Inter-residue/atomic distance constraints（contact）
- Pocket distance constraints（pocket）
Note that constraints are not strictly enforced—the final output structure is determined by the prediction algorithm.
For more detailed information on constraints, please refer to the official documentation: https://github.com/chaidiscovery/chai-lab/blob/main/examples/restraints/README.md
3. Result Interpretation
The output will include 5 predicted .cif structure files, along with a corresponding prediction metrics .csv file for each structure. Below is an explanation of the core metrics:
| Metric | Full Name | Value Range | Focus Level | Typical Threshold/Interpretation | Main Purpose |
|---|---|---|---|---|---|
| Aggregate score | Aggregate confidence score | 0–1 | Whole complex | 0.2 * overall PTM score + 0.8 * ipTM score - 100 * if there is chain conflict | Select high-quality results |
| pTM | Predicted Template Modeling score | 0–1 | Single chain / whole complex | ≥ 0.5 indicates basic correct fold | Global Topology Reliability |
| ipTM | Interface Predicted TM-score | 0–1 | Interface between chains | ≥ 0.8 indicates high-quality docking | Inter-chain interactions in complexes |
4. References
[1] Chai Discovery, Jacques Boitreaud, Jack Dent, Matthew McPartlon, Joshua Meier, Vinicius Reis, Alex Rogozhnikov, Kevin Wu. Chai-1: Decoding the molecular interactions of life. bioRxiv 2024.10.10.615955. https://doi.org/10.1101/2024.10.10.615955
[2] Mirdita, M., Schütze, K., Moriwaki, Y. et al. ColabFold: making protein folding accessible to all. Nat Methods 19, 679–682 (2022). https://doi.org/10.1038/s41592-022-01488-1