Ken Dill (UCSF) and Vojko Vlachy (Slovenia) are collaborating to develop physics-based models of water and aqueous solvation. Currently, there are two choices for computational modeling of solvent: (1) Explicit solvent models - TIP, SPC, etc., which are computationally too expensive for many applications. (2) Implicit solvent models - Poisson-Boltzmann (PB) or Generalized Born (GB), which replace water molecules with a continuum. PB or GB models often give incorrect free energies because they miss the hydrogen bonding, the particulate nature of water, and effects of surface geometry. Here, we propose a third approach: we are treating water molecules and hydrogen bonding explicitly, but with analytical methods - integral equations, thermodynamic perturbation theory, density functional theory, and mean-field models - that are computationally efficient. In preliminary work, these methods are already 3 orders of magnitude faster to compute, and they correct several of the flaws of simpler models. If successful, the developments here would lead to better ways to handle solvation in computational biology.
Showing the most recent 10 out of 74 publications
