RGI-Toolkit: Differentiable Restraints for Controllable Biomolecular Structure Prediction
Diffusion-based models have substantially advanced biomolecular structure prediction. Within this framework, restraint-guided inference (RGI) corrects the denoiser output at each reverse-diffusion step using a differentiable restraint loss, allowing available experimental information, stereochemical requirements, or a specified target conformational state to be incorporated into predictions…
Diffusion-based models have significantly improved the prediction of biomolecular structures. Restrain-guided inference (RGI) is a technique used to refine these predictions by incorporating experimental data, stereochemical constraints, or a desired target conformational state into the model's output during each reverse-diffusion process.
RGI-Toolkit, a Python library, enables the use of RGI with various structure predictors. This toolkit streamlines the process of defining restraints, selecting atoms, evaluating loss functions, and optimizing coordinates within a single, adaptable engine. Restraints can be applied to distance, angle, and dihedral parameters, as well as deviations from reference structures and user-defined reaction coordinates.
Ligand conformer restraints are also supported. The toolkit currently integrates with six predictors, including AlphaFold3. By utilizing restraints, particularly for conformation and stereochemical properties, the toolkit reduces the occurrence of chirality and cis/trans errors in predicted ligands and guides the protein conformations towards the desired target states.
Written by urgent.news from bioRxiv's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
