Statistically Optimal Force Aggregation for Coarse-Graining Molecular Dynamics.

Autor:	Krämer A; Department of Mathematics and Computer Science, Freie Universität Berlin, Arnimallee 12, 14195 Berlin, Germany., Durumeric AEP; Department of Mathematics and Computer Science, Freie Universität Berlin, Arnimallee 12, 14195 Berlin, Germany., Charron NE; Department of Physics and Astronomy, Rice University, Houston, Texas 77005, United States.; Center for Theoretical Biological Physics, Rice University, Houston, Texas 77251, United States.; Department of Physics, Freie Universität Berlin, Arnimallee 12, 14195 Berlin, Germany., Chen Y; Department of Mathematics and Computer Science, Freie Universität Berlin, Arnimallee 12, 14195 Berlin, Germany.; International Max Planck Research School for Biology and Computation (IMPRS-BAC), Max Planck Institute for Molecular Genetics, 14195 Berlin, Germany., Clementi C; Department of Physics and Astronomy, Rice University, Houston, Texas 77005, United States.; Center for Theoretical Biological Physics, Rice University, Houston, Texas 77251, United States.; Department of Physics, Freie Universität Berlin, Arnimallee 12, 14195 Berlin, Germany.; Department of Chemistry, Rice University, Houston, Texas 77005, United States., Noé F; Department of Mathematics and Computer Science, Freie Universität Berlin, Arnimallee 12, 14195 Berlin, Germany.; Department of Physics, Freie Universität Berlin, Arnimallee 12, 14195 Berlin, Germany.; Department of Chemistry, Rice University, Houston, Texas 77005, United States.; Microsoft Research AI4Science, Karl-Liebknecht Straße 32, 10178 Berlin, Germany.
Jazyk:	angličtina
Zdroj:	The journal of physical chemistry letters [J Phys Chem Lett] 2023 May 04; Vol. 14 (17), pp. 3970-3979. Date of Electronic Publication: 2023 Apr 20.
DOI:	10.1021/acs.jpclett.3c00444
Abstrakt:	Machine-learned coarse-grained (CG) models have the potential for simulating large molecular complexes beyond what is possible with atomistic molecular dynamics. However, training accurate CG models remains a challenge. A widely used methodology for learning bottom-up CG force fields maps forces from all-atom molecular dynamics to the CG representation and matches them with a CG force field on average. We show that there is flexibility in how to map all-atom forces to the CG representation and that the most commonly used mapping methods are statistically inefficient and potentially even incorrect in the presence of constraints in the all-atom simulation. We define an optimization statement for force mappings and demonstrate that substantially improved CG force fields can be learned from the same simulation data when using optimized force maps. The method is demonstrated on the miniproteins chignolin and tryptophan cage and published as open-source code.
Databáze:	MEDLINE
Externí odkaz:	Zobrazit plný text záznamu