In order to simulate the interactions of molecules, an accurate method of representing these interactions must be available. Ab initio calculations are able to do this, but these calculations are restricted to relatively small systems. The simulation of larger systems requires a simpler, less computationally intensive method such as molecular mechanics (MM). The force fields employed in MM and MD calculations are a much simpler representation of the potential energy of the system compared to ab initio methods. Fortunately, these force fields have been shown to accurately model many types of systems. Because solvation effects are very important in many biological processes, e.g. catalysis and binding, it is also necessary to model these systems in the condensed phase with the ability to accurately calculate the energetics of desolvation. In recent years, free energy perturbation (FEP) calculations have reproduced experimental values for the free energy of solvation of a variety of molecules, including those with polar and hydrophobic groups. The ability to correlate calculated results to experimental data allows the calibration of the potential functions which are employed. With accurate potential functions, these methods can be used in a predictive manner for molecules whose solvation energies are not accessible experimentally. Cornell et al. have recently re-parameterized the AMBER force field. I am interested in using this new force field to study molecules whose solvation free energies have not been determined experimentally. The nucleic acid bases are one such class of molecules. It is important to be able to accurately model these molecules because the formation of duplex or triplex nucleic acids depends on the energetics of desolvating the bases. The free energies of solvation of 9-methyladenine and 1-methylthymine have been derived from solubilities reported by Clark et al. These free energies have also been calculated by several methods: FEP with the old AMBER 4.0 force field; AMSOL, a recently developed semi-empirical model that includes solvation parameters; and FEP with the OPLS parameter set. The results obtained from the three models and parameter sets are very different. These differences bring into question the accuracy of the models, especially in light of the excellent agreement with experimental data for these models for other classes of polar and hydrophobic molecules. I plan to test the new force field by calculating the solvation free energies of adenine and thymine, and use it in a predictive manner by calculating the energies for guanine and cytosine. For this project I use the Computer Graphics Laboratory facilities in the following ways: MidasPlus to examine the geometry of all structures at the beginning and ending points of the perturbation calculations. In addition, I use the image generation tools when preparing posters and publications.
Showing the most recent 10 out of 508 publications
