UC San Diego

Biomolecular recognition, the nexus of individual molecules and function, and solvation, the medium in which it occurs, are two of the most important, well-studied, and yet enigmatic phenomena in biology. This thesis explores the combination of theory and computational algorithms to study these two phenomena. End-point free energy calculations are used to predict and characterize biomolecular recognition. A theoretical framework is presented which properly accounts for standard state conditions and highlights several areas of potential improvement. A novel method is presented for calculating the association free energy, which arises from one molecule losing its translational and rotational freedom upon binding. The entropic penalties of protein-ligand and protein-protein associations are compared and the physical trends that affect this penalty are presented. Compatibility between the implicit and explicit solvent energies is shown to be especially important in end-point free energy analysis. Energetic compatibility, in addition to implicit solvent accuracy, can be improved by fitting implicit solvent parameters to explicit solvent simulations. Several variations of this approach are presented, each of which optimizes the solute radii which define the dielectric boundary between the low dielectric solute and the high dielectric solvent in Poisson-based implicit solvent models. The radii are first optimized to reproduce explicit solvent charging free energies from explicit solvent simulations. Then, atomic forces are shown to increase the optimization efficiency and improve the resulting implicit solvent parameters. Molecular surfaces are compared to atom-centered dielectric functions, and the latter are shown to create unphysical high dielectric regions in spaces between atoms which are too small for water molecules to penetrate. Optimized radii are presented for molecular surfaces and spline- smoothed surfaces created for stable and efficient force calculations. Finally, the coupling between polar and nonpolar solvation energies is discussed. This coupling has never been accounted for in an implicit solvent framework, but has been demonstrated in solutes ranging from small molecules to nanosolutes, including several biological systems. A new implicit solvent formalism is presented which accounts for coupling by expressing the system free energy as a functional of the solvent volume exclusion function