A Direct Method for Incorporating Experimental Data into Multiscale Coarse-Grained Models.

Area of Science:

• Computational chemistry
• Statistical mechanics
• Molecular modeling

Background:

• Extracting meaningful data from molecular simulations requires integrating new experimental observations.
• Experiment-directed simulation (EDS) is a recent method using maximum entropy to bias models towards experimental data with minimal changes.
• The systematic improvement of models via minimal bias in EDS has not been previously discussed.

Purpose of the Study:

• To demonstrate that minimal bias, as used in EDS, systematically reduces the relative entropy of a system.
• To show that EDS can rapidly improve bottom-up multiscale coarse-grained (MS-CG) models without extensive reparametrization.
• To establish a new paradigm combining bottom-up and top-down approaches in coarse-grained modeling.

Main Methods:

• Applying minimal bias using experiment-directed simulation (EDS) to all-atom simulations.
• Analyzing the change in relative entropy to quantify the improvement from minimal bias.
• Projecting many-body interactions from EDS bias onto effective two-body MS-CG interactions.

Main Results:

• The relative entropy of a biased system is always reduced by the application of a minimal bias like that in EDS.
• EDS allows for rapid improvement of MS-CG models without time-consuming reparametrization of atomistic force fields.
• Many-body interactions introduced by EDS bias can be effectively maintained in projected two-body MS-CG interactions.

Conclusions:

• EDS provides a rigorous statistical mechanics framework for improving molecular models.
• This work introduces a novel paradigm combining bottom-up and top-down strategies in coarse-grained modeling.
• The EDS-MS-CG models demonstrate utility in simulating molecular systems like liquid methanol and ethylene carbonate.