Molecular complexes formed by proteins and small-molecule ligands are ubiquitous, and predicting their 3D structures can facilitate both biological discoveries and the design of novel enzymes or drug molecules. We propose a deep generative model framework to rapidly predict protein-ligand complex structures and their fluctuations. The NeuralPLexer framework jointly models protein and small-molecule coordinates at an atomistic resolution through a time-truncated diffusion process that incorporates biophysical constraints and inferred proximity information. The generative diffusion process is learned by a novel stereochemistry-aware equivariant graph transformer that enables efficient and scalable gradient field prediction for all heavy atoms in the protein-ligand complex. NeuralPLexer outperforms existing physics-based and learning-based methods on benchmarking problems including blind protein-ligand docking and ligand-coupled binding site design. Moreover, NeuralPLexer enriches bound-state-like protein structures when applied to systems where protein folding landscapes are significantly altered by the presence of ligands. Our results reveal that a data-driven approach can capture the structural cooperativity among protein and small-molecule entities, showing promise for the computational identification of novel drug targets and the end-to-end differentiable design of functional small-molecules and ligand-binding proteins.
Zhuoran Qiao is currently a Lead Machine Learning Scientist at Entos, Inc. He recently graduated from the California Institute of Technology with a PhD degree in chemistry. His PhD study was advised by Professor Anima Anandkumar, and was partially supported by the Amazon/Caltech AI4Science Fellowship. His research centres around physics-inspired machine learning approaches to tackle complex problems in chemistry and biology, inclduing electronic-structure-based ML methods for molecular modelling, differentiable quantum chemistry methods, and deep generative models for biomolecular structure prediction.
To become a member of the Rough Path Interest Group, register here for free.