Machine learning coarse-grained potentials of protein thermodynamics

被引：33

作者：

Majewski, Maciej ^{[1
,2
]}

Perez, Adria ^{[1
,2
]}

Tholke, Philipp ^{[1
]}

Doerr, Stefan ^{[2
]}

Charron, Nicholas E. ^{[3
,4
,5
]}

Giorgino, Toni ^{[6
]}

Husic, Brooke E. ^{[7
,8
,9
,10
]}

Clementi, Cecilia ^{[3
,4
,5
,11
]}

Noe, Frank ^{[5
,7
,11
,12
]}

De Fabritiis, Gianni ^{[1
,2
,13
]}

机构：

[1] Univ Pompeu Fabra, Sci Computat Lab, Biomed Res Pk PRBB,Carrer Dr Aiguader 88, Barcelona 08003, Spain

[2] Acellera Labs, Doctor Trueta 183, Barcelona 08005, Spain

[3] Rice Univ, Dept Phys, Houston, TX 77005 USA

[4] Rice Univ, Ctr Theoret Biol Phys, Houston, TX 77005 USA

[5] FU Berlin, Dept Phys, Arnimallee 12, D-14195 Berlin, Germany

[6] Natl Res Council CNR IBF, Inst Biophys, I-20133 Milan, Italy

[7] FU Berlin, Dept Mathe & Comp Sci, Arnimallee 12, D-14195 Berlin, Germany

[8] Princeton Univ, Lewis Sigler Inst Integrat Genom, Princeton, NJ 08540 USA

[9] Princeton Univ, Princeton Ctr Theoret Sci, Princeton, NJ 08540 USA

[10] Princeton Univ, Ctr Phys Biol Funct, Princeton, NJ 08540 USA

[11] Rice Univ, Dept Chem, Houston, TX 77005 USA

[12] Microsoft Res AI4Sci, Karl Liebknecht Str 32, D-10178 Berlin, Germany

[13] Inst Catalana Recerca & Estudis Avancats ICR, Passeig Lluis Companys 23, Barcelona 08010, Spain

来源：

NATURE COMMUNICATIONS | 2023年 / 14卷 / 01期

基金：

美国国家科学基金会; 美国国家卫生研究院; 欧洲研究理事会; 欧盟地平线“2020”;

关键词：

MOLECULAR-DYNAMICS SIMULATIONS; FORCE-FIELD; STRUCTURE PREDICTION; ENERGY LANDSCAPES; STATE MODELS; PERSPECTIVE; KINETICS;

D O I：

10.1038/s41467-023-41343-1

中图分类号：

O [数理科学和化学]; P [天文学、地球科学]; Q [生物科学]; N [自然科学总论];

学科分类号：

07 ; 0710 ; 09 ;

摘要：

A generalized understanding of protein dynamics is an unsolved scientific problem, the solution of which is critical to the interpretation of the structure-function relationships that govern essential biological processes. Here, we approach this problem by constructing coarse-grained molecular potentials based on artificial neural networks and grounded in statistical mechanics. For training, we build a unique dataset of unbiased all-atom molecular dynamics simulations of approximately 9 ms for twelve different proteins with multiple secondary structure arrangements. The coarse-grained models are capable of accelerating the dynamics by more than three orders of magnitude while preserving the thermodynamics of the systems. Coarse-grained simulations identify relevant structural states in the ensemble with comparable energetics to the all-atom systems. Furthermore, we show that a single coarse-grained potential can integrate all twelve proteins and can capture experimental structural features of mutated proteins. These results indicate that machine learning coarse-grained potentials could provide a feasible approach to simulate and understand protein dynamics.

引用

页数：13

共 97 条

[1] Accurate prediction of protein structures and interactions using a three-track neural network [J].