Page 1 of 1
Integration of machine learning into molecular simulations
Posted: Tue Mar 01, 2022 12:52 pm
by JimF
NNP/MM is a hybrid simulation method combining neural network potentials (NNP) and molecular mechanics (MM). NNP can model the molecular interactions more accurately than the conventional force fields in MM, but it still is not as fast as MM. Thus, only the important part of a molecular system is simulated with NNP, while the rest part is using MM.
http://www.gpugrid.net/forum_thread.php?id=5305
(1) Is this relevant to Folding?
(2) Will it be incorporated in some future OpenMM version?
Re: Integration of machine learning into molecular simulatio
Posted: Tue Mar 08, 2022 5:30 pm
by toTOW
I got an answer from Chodera Lab ...
Alex Payne wrote:The short answer is that, yes, people in the Chodera lab have definitely looking into some of these methods, but the implementation into F@h I don't know about.
And two related publications :
Teaching free energy calculations to learn from experimental data
Towards chemical accuracy for alchemical free energy calculations with hybrid physics-based machine learning / molecular mechanics potentials
Re: Integration of machine learning into molecular simulatio
Posted: Tue Mar 08, 2022 8:02 pm
by toTOW
John Chodera wrote:We are working on it! Once OpenMM 8 is released, we will be able to integrate NNP/MM (or QML/MM, whatever you want to call it) into core22! We’re just working out the final packaging / library issues right now. Practically speaking, it’s just a matter of pulling in the plugins to the core22 build once we’ve released OpenMM 8 and friends. So yes, it’s on our roadmap, coming shortly after OpenMM 8 release.