AbstractDifferential geometry (DG) based solvation models have shown their great success in solvation analysis by avoiding the use of ad hoc surface definitions, coupling the polar and nonpolar free energies, and generating solvent-solute boundary in a physically self-consistent fashion. Parameter optimization is a key factor for their accuracy, predictive ability of solvation free energies, and other applications. Recently, a series of efforts have been made to improve the parameterization of these new implicit solvent models. In thiswork, we aim at studying the role of dispersion attraction in the parameterization of our DG based solvation models. To this end, we first investigate the necessity of van derWaals (vdW) dispersion interactions in the model and then carry out systematic parameterization for the model in the absence of electrostatic interactions. In particular, we explore how the changes in Lennard-Jones (L-J) potential expression, its decomposition scheme, and choices of some fixed parameter values affect the optimal values of other parameters as well as the overall modeling error. Our study on nonpolar solvation analysis offers insights into the parameterization of nonpolar components for the full DG based models by eliminating uncertainties from the electrostatic polar component. Therefore, it can be regarded as a step towards better parameterization for the full DG based model.
Role of Dispersion Attraction in Differential Geometry Based Nonpolar Solvation Models
