AMARO: All Heavy-Atom Transferable Neural Network Potentials of Protein Thermodynamics

TL;DR

AMARO is a novel neural network potential that enables efficient, scalable, and generalizable all-atom protein simulations by combining an O(3)-equivariant architecture with coarse-graining, reducing computational costs.

Contribution

It introduces AMARO, a new neural network potential that effectively models protein thermodynamics with a coarse-grained approach and no prior energy terms.

Findings

AMARO enables stable protein dynamics simulations.

It demonstrates scalability and generalization in protein modeling.

The method reduces computational costs compared to traditional approaches.

Abstract

All-atom molecular simulations offer detailed insights into macromolecular phenomena, but their substantial computational cost hinders the exploration of complex biological processes. We introduce Advanced Machine-learning Atomic Representation Omni-force-field (AMARO), a new neural network potential (NNP) that combines an O(3)-equivariant message-passing neural network architecture, TensorNet, with a coarse-graining map that excludes hydrogen atoms. AMARO demonstrates the feasibility of training coarser NNP, without prior energy terms, to run stable protein dynamics with scalability and generalization capabilities.