Predicting 3D structures of biomolecules has become an essential step in molecular modeling and designing ligands, drugs, or understanding biological mechanisms. However, one common pain point faced by many researchers is how to incorporate prior biological knowledge—like known binding sites or residue-residue interactions—into structure prediction workflows.
This is where Chai-1, available through the Biomolecular Structure Prediction extension in SAMSON, offers a useful feature: the ability to apply spatial restraints during prediction. This is particularly useful when you want to model a biomolecule in a way that reflects known or expected interactions.
What Are Restraints in Structure Prediction?
Restraints are constraints you provide to the prediction algorithm, prompting it to keep certain atoms or residues close (or at a known distance) from each other. In Chai-1, two types are available:
- Pocket restraints: Limit the distance between a residue and a chain. Useful to guide potential ligand-binding pockets placement.
- Contact restraints: Enforce a distance between two specific residues. This is useful when NMR data or co-evolutionary analyses suggest contacts you want the structure to respect.
How to Use Chai-1 with Restraints
Here’s a quick guide to adding structure refinement with restraints via Chai-1 in SAMSON:
- Open Home > Predict in SAMSON.
- Select the Chai-1 service.
- Add your input sequences using Add protein, Add DNA, Add RNA, or Add ligand. Input formats include SMILES strings and sequences.
- To spatially guide your prediction:
- Use Add pocket restraint to define interactions between a residue and a chain (e.g., proximity to guide ligand placement).
- Use Add contact restraint to define distance constraints between two residues.
- Click Start prediction.
Predictions are run on cloud-based A100 GPUs, with a typical job costing between 0.5 and 1 computing credit. These cloud resources provide the performance necessary for incorporating such restraints effectively.
Why Use Restraints?
Restraints offer several benefits:
- Allow you to incorporate experimental data (e.g., crosslinking, mutagenesis, or NMR-based distance predictions).
- Improves prediction realism in the context of known residue interactions or binding sites.
- Reduces computational time by limiting solution space.
This can be particularly useful in complex systems such as protein-DNA complexes, multi-domain proteins, or cases where ligand-binding geometries matter for downstream modeling, like virtual screening or docking.
Unlike some structure prediction tools that treat proteins as black boxes, Chai-1 accommodates user-driven guidance, making it well-suited for integrative structural biology workflows.
While Chai-1 currently supports restraints between residues and between residues and chains, it would be beneficial to keep an eye out for future versions that might expand possibilities (e.g., incorporating chemical shift-based guidance or constraints from cryo-EM maps).
Learn More
To dive deeper into Chai-1 and how it fits into SAMSON’s ecosystem for molecular design, see the full documentation page here: https://documentation.samson-connect.net/tutorials/bsp/bsp/.
SAMSON and all SAMSON Extensions are free for non-commercial use. You can download SAMSON at https://www.samson-connect.net.