Constructing Predictive Implicit Solvent Models: Insights from Exploring Potential Parameter Space and Property Predictions.

Abstract

Implicit solvent (IS) models enhance the efficiency of simulating liquid systems, but those based on the potential of mean force calculations using explicit solvent (ES) models often fail to capture the solute structure and assembly dynamics under various conditions. This study examines the relationships between the parameter space of IS models and their predictive capabilities regarding solute structure and assembly dynamics, focusing on NaCl solutions and protein-bound silica colloidal nanoparticles. For NaCl solutions, models developed using concentration-dependent dielectric constants generally lack efficacy, and their effectiveness in predicting high-concentration liquid structures varies based on the ES model used. Adjusting the dielectric constant value or the shape of the short-range potentials can improve the accuracy of these models in predicting liquid structures. For protein-bound silica colloidal nanoparticles, we find that interaction site size and model resolution significantly influence the prediction of nanoparticle assembly dynamics, with different resolutions producing aggregates with distinct structures and connection modes. This study elucidates the complex relationship between parameter space in IS models and liquid structure and assembly kinetics, providing crucial insights for developing more accurate and predictive models.