jwohlwend/boltz
Boltz
Boltz is a deep learning molecular modeler and biomolecular structure prediction system. It uses neural network architectures to simulate the physical folding and docking of biomolecules, specifically predicting the three-dimensional shapes of protein and ligand complexes.
The project functions as a protein-ligand complex predictor and binding affinity predictor, estimating the strength and probability of molecular interactions between ligands and targets. These capabilities are applied to computer aided drug design, including ligand binding affinity prediction and protein-ligand interaction modeling.
The system incorporates coordinate refinement and geometric modeling to determine atomic positions, alongside sequence embedding and folding simulations to predict stable molecular conformations.
Features
- Structure Prediction Models - Predicts three-dimensional biomolecular coordinates by learning spatial relationships and folding patterns from protein sequence data.
- Binding Affinity Prediction - Estimates the probability that a ligand binds to a target or calculates specific affinity values for lead compound optimization.
- Computational Workbenches - Provides a computational toolkit for predicting molecular interactions and structures to accelerate therapeutic compound development.
- Molecular Deep Learning Frameworks - Utilizes a neural network architecture designed to simulate the physical folding and docking of biomolecules.
- Protein Interaction Modeling - Analyzes how small molecules interact with protein targets to predict binding sites and effectiveness of drug candidates.
- Affinity Scoring Functions - Calculates binding probability and strength by comparing energetic states of bound and unbound molecular complexes.
- Geometric Modeling - Implements attention-based geometric modeling to determine the relative distance and orientation of atoms.
- Coordinate Refinement - Iteratively denoises coarse structural predictions to achieve high-resolution atomic coordinates using a stochastic generative process.
- Sequence Featurization - Transforms raw protein sequences into high-dimensional feature vectors using evolutionary information from related sequences.
- Cotranslated Folding Simulations - Models the process of protein folding during synthesis to predict the final stable molecular conformation.
