# A multi-grained symmetric differential equation model for learning protein-ligand binding dynamics

## TL;DR

NeuralMD is a new AI framework that improves the accuracy and efficiency of simulating how proteins and ligands bind, which could speed up drug discovery.

## Contribution

NeuralMD introduces a physics-informed, multi-grained, group-symmetric framework for modeling protein-ligand binding dynamics.

## Key findings

- NeuralMD achieves up to 15× lower reconstruction error compared to existing ML baselines.
- The model shows 70% higher validity in predicting protein-ligand binding dynamics.
- Predicted oscillations closely align with ground-truth dynamics in simulations.

## Abstract

Molecular dynamics (MD) simulation is a key tool in drug discovery for predicting protein-ligand binding affinities, transport properties, and pocket dynamics. While advances in numerical and machine learning (ML) methods have improved MD efficiency, accurately modeling long-timescale dynamics remains challenging. We introduce NeuralMD, an ML surrogate that accelerates and enhances MD simulations of protein-ligand binding. NeuralMD employs a physics-informed, multi-grained, group-symmetric framework comprising (1) BindingNet, which enforces symmetry via vector frames and captures multi-level protein-ligand interactions, and (2) an augmented neural differential equation solver that learns trajectories under Newtonian mechanics. Across ten single-trajectory and three multi-trajectory tasks, NeuralMD achieves up to 15 × lower reconstruction error and 70% higher validity than existing ML baselines. The predicted oscillations closely align with ground-truth dynamics, establishing NeuralMD as a foundation for next-generation protein-ligand simulation research.

Artificial intelligence is advancing molecular simulation. Here, the authors introduce NeuralMD, a physics-informed AI framework that efficiently models and predicts protein-ligand binding dynamics, providing opportunities for accelerating drug discovery.