OpenMM 8: Molecular Dynamics Simulation with Machine Learning Potentials

被引：109

作者：

Eastman, Peter ^{[1
]}

Galvelis, Raimondas ^{[2
,3
]}

Pelaez, Raul P. ^{[3
]}

Abreu, Charlles R. A. ^{[4
,5
]}

Farr, Stephen E. ^{[6
]}

Gallicchio, Emilio ^{[7
,8
,9
]}

Gorenko, Anton ^{[10
]}

Henry, Michael M. ^{[11
]}

Hu, Frank ^{[1
]}

Huang, Jing ^{[12
]}

Kramer, Andreas ^{[13
]}

Michel, Julien ^{[6
]}

Mitchell, Joshua A. ^{[14
]}

Pande, Vijay S. ^{[15
,16
]}

Rodrigues, Joao P. G. L. M. ^{[16
]}

Rodriguez-Guerra, Jaime ^{[17
]}

Simmonett, Andrew C. ^{[18
]}

Singh, Sukrit ^{[11
]}

Swails, Jason ^{[19
]}

Turner, Philip ^{[20
]}

Wang, Yuanqing ^{[21
,22
]}

Zhang, Ivy ^{[11
,23
]}

Chodera, John D. ^{[11
]}

De Fabritiis, Gianni ^{[2
,3
,24
]}

Markland, Thomas E. ^{[1
]}

机构：

[1] Stanford Univ, Dept Chem, Stanford, CA 94305 USA

[2] Acellera Labs, Barcelona 08005, Spain

[3] Univ Pompeu Fabra, Computat Sci Lab, Barcelona 08003, Spain

[4] Univ Fed Rio De Janeiro, Sch Chem, Chem Engn Dept, BR-68542 Rio De Janeiro, Brazil

[5] Redesign Sci Inc, New York, NY 10014 USA

[6] Univ Edinburgh, EaStCHEM Sch Chem, Edinburgh EH9 3FJ, Scotland

[7] CUNY Brooklyn Coll, Dept Chem & Biochem, Brooklyn, NY 11210 USA

[8] CUNY, Grad Ctr, PhD Program Chem, New York, NY 10016 USA

[9] CUNY, PhD Program Biochem, Grad Ctr, New York, NY 10016 USA

[10] Stream HPC, NL-1062 HG Amsterdam, Netherlands

[11] Mem Sloan Kettering Canc Ctr, Sloan Kettering Inst, Computat & Syst Biol Program, New York, NY 10065 USA

[12] Westlake Univ, Sch Life Sci, Key Lab Struct Biol Zhejiang Prov, Hangzhou 310024, Zhejiang, Peoples R China

[13] Free Univ Berlin, Freie Universitat Berlin, D-14195 Berlin, Germany

[14] Open Mol Software Fdn, Open Force Field Initiat, Davis, CA 95616 USA

[15] Andreessen Horowitz, Menlo Pk, CA 94025 USA

[16] Stanford Univ, Dept Struct Biol, Stanford, CA 94305 USA

[17] Charitee Univ Med Berlin, In Silico Toxicol & Struct Bioinformat, D-10117 Berlin, Germany

[18] NIH, Lab Computat Biol, NHLBI, Bethesda, MD 20892 USA

[19] Entos Inc, La Jolla, CA 92037 USA

[20] Virginia Polytech Inst & State Univ, Coll Engn, Blacksburg, VA 24061 USA

[21] NYU, Simons Ctr Computat Phys Chem, New York, NY 10004 USA

[22] NYU, Ctr Data Sci, New York, NY 10004 USA

[23] Cornell Univ, Weill Cornell Med Coll, Triinst PhD Program Computat Biol & Med, New York, NY 10065 USA

[24] ICREA, Barcelona 08010, Spain

来源：

JOURNAL OF PHYSICAL CHEMISTRY B | 2023年 / 128卷 / 01期

基金：

美国国家卫生研究院; 英国工程与自然科学研究理事会;

关键词：

GENERALIZED GRADIENT APPROXIMATION; FORCE-FIELD; ACCURACY;

D O I：

10.1021/acs.jpcb.3c06662

中图分类号：

O64 [物理化学（理论化学）、化学物理学];

学科分类号：

070304 ; 081704 ;

摘要：

Machine learning plays an important and growing role in molecular simulation. The newest version of the OpenMM molecular dynamics toolkit introduces new features to support the use of machine learning potentials. Arbitrary PyTorch models can be added to a simulation and used to compute forces and energy. A higher-level interface allows users to easily model their molecules of interest with general purpose, pretrained potential functions. A collection of optimized CUDA kernels and custom PyTorch operations greatly improves the speed of simulations. We demonstrate these features in simulations of cyclin-dependent kinase 8 (CDK8) and the green fluorescent protein chromophore in water. Taken together, these features make it practical to use machine learning to improve the accuracy of simulations with only a modest increase in cost.

引用

页码：109 / 116

页数：8

共 51 条

[1] DFTB+, a sparse matrix-based implementation of the DFTB method [J].