A Fast and Accurate Analytical Method for the Computation of Solvent Effects in Molecular Simulations

Citation

Zamanakos, Georgios (2002) A Fast and Accurate Analytical Method for the Computation of Solvent Effects in Molecular Simulations. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/B7W8-N760. https://resolver.caltech.edu/CaltechETD:etd-04062005-082441

Abstract

The solvent environment of molecules plays a very important role in their structure and function. In biological systems it is well known that water has profound effects in the functions of proteins. Simulations assist us in microscopic studies of chemical and biological phenomena. It is important then to include solvation effects accurately and efficiently in molecular simulations. In this work we present a novel approximate analytical method for calculating the solvation energy for every atom of a molecular system and the forces that act on each atom because of the solvent. The solvation energy is partitioned into long-range and short-range contributions. The longrange contributions are due to polar interactions between the solvent and the solute and the short-range are due to van der Waals and entropic effects. We show how the calculation of these effects, under certain approximations, can be reduced to the calculation of the volume and exposed area of each atom, assuming a fused-sphere model for the solute. We demonstrate a fast method for the exact, analytical calculation of the volume and area of each atom in the fused-sphere model and their gradients with respect to the atom's position. We incorporate the fast geometric algorithms into the approximate formulas we derived for the calculation of the solvation energy, to get our solvation model, the Analytical Volume Generalized Born - Solvent Accessible Surface (AVGBSAS) model.

The predictions of the polar part of the method (AVGB) are very good as compared to numerical solutions of the underlying physical model, the Poisson-Boltzman equation, for small and large molecular systems. AVGB does not depend on any fitting parameters, which is common in the literature for such approximate methods. It is very fast compared to numerical solutions of the PB equation or other Generalized Born methods. Also, the method is parallelizable which allows us to study much larger systems. The AVGB-SAS method has been implemented in a parallel molecular dynamics software package and a molecular docking software package. We have demonstrated the quality of the results of the AVGB-SAS model in the dynamics of DNA and in rational drug design applications.

Thesis Files