Automatically Constructed Neural Network Potentials for Molecular Dynamics Simulation of Zinc Proteins

Abstract

The development of accurate and efficient potential energy functions for the molecular dynamics simulation of metalloproteins has long been a great challenge for the theoretical chemistry community. An artificial neural network provides the possibility to develop potential energy functions with both the efficiency of the classical force fields and the accuracy of the quantum chemical methods. In this work, neural network potentials were automatically constructed by using the ESOINN-DP method for typical zinc proteins. For the four most common zinc coordination modes in proteins, the potential energy, atomic forces, and atomic charges predicted by neural network models show great agreement with quantum mechanics calculations and the neural network potential can maintain the coordination geometry correctly. In addition, MD simulation and energy optimization with the neural network potential can be readily used for structural refinement. The neural network potential is not limited by the function form and complex parameterization process, and important quantum effects such as polarization and charge transfer can be accurately considered. The algorithm proposed in this work can also be directly applied to proteins containing other metal ions.

Drug Design in the Exascale Era: A Perspective from Massively Parallel QM/MM Simulations

“…Because of the cost associated with exascale calculations, we expect DFT QM/MM MD calculations to tremendously profit from the diffusion of ML techniques in molecular simulations . Indeed, hybrid ML/MM models enable the simulation of biological systems using an ML representation of a quantum mechanical potential at near QM/MM accuracy and at a fraction of the computational cost. ,− These ML models work natively on GPUs, and because they normally rely on local interactions alone, they can be exceptionally scalable on distributed architectures. , Furthermore, their training requires data sets generated through many single-point QM­(/MM) calculations that are expensive but embarrassingly parallelizable. Finally, the recent introduction of ML-accelerated perturbative techniques provides an efficient and highly parallelizable way of recovering the accuracy of QM/MM potentials from simulations using cheaper methods (such as force fields or even ML/MM models) at the cost of only a few single-point energy and force QM/MM calculations. ,, These methods, in combination with enhanced sampling approaches, promise to enable the QM/MM prediction of fundamental biophysical quantities such as drug–protein binding free energies or full free energy surfaces.…”

Drug Design in the Exascale Era: A Perspective from Massively Parallel QM/MM Simulations

“…123 Indeed, hybrid ML/MM models enable the simulation of biological systems using an ML representation of a quantum mechanical potential at near QM/MM accuracy and at a fraction of the computational cost. 44,[124][125][126][127] These ML models work natively on GPUs, and because they normally rely on local interactions alone, they can be exceptionally scalable on distributed architectures. 72,73 Furthermore, their training requires datasets generated through many single-point QM(/MM) calculations that are expensive but embarrassingly parallelizable.…”

A fast–slow method to treat solute dynamics in explicit solvent

“…The implicit solvent employed the GB model with “igb = 8” ( GBneck2 ). 50 For the explicit solvent model, the spherical water box with a completely rigid boundary was used as a solvent environment, 51 and the buffer distance between the solute and the boundary was 12 Å. The three test peptides/proteins were run for three 10 ns and one 100 ns simulation in the NVT ensemble without any constraints.…”