Abstract

Continuum models for molecular solvation, such as the venerable Poisson-Boltzmann equation, provide valuable capabilities for studies in which explicit solvent simulations are intractable or impractical. Unfortunately, the fidelity of existing continuum models is limited due to the substantial physical approximations inherent to modeling solvent implicitly. In this talk, I will describe our development of advanced implicit-solvent models that leverage important methods in computational engineering and materials modeling. First, we are using multiscale continuum theory to incorporate length-scale dependent phenomena—in contrast to standard Poisson-based models, which are scale invariant. Second, we have shown that nonlinear boundary conditions at the molecule-solvent interface enable the first accurate continuum calculations of charge-sign dependent asymmetric solvation. Third, we use fast simulation software based on boundary-integral equation (BIE) formulations of the more widely used partial-differential equation (PDE) approach—for example, the software applications DelPhi and APBS . Our BIE approach provides not only straightforward parallel scaling for large-scale simulations with millions of solute atoms, but also favorable mathematical properties enabling an improved, more mathematically rigorous version of Generalized-Born (GB) theory that offers tunable accuracy.