Heuristic molecular lipophilicity potential for computer-aided rational drug design

Abstract

In my thesis research, I suggest a heuristic molecular lipophilicity potential (HMLP), a structure-based technique requiring no empirical indices of atomic lipophilicity, for computer-aided drug design. The input data used in this approach are molecular geometries and molecular surfaces. The HMLP is a modified electrostatic potential, combined with the averaged influences from the molecular environment. Quantum mechanics is used in calculating the electron density function ñ(r) and the electrostatic potential V(r), and from this information a lipophilicity potential L(r) is generated. The HMLP is a unified lipophilicity and hydrophilicity potential. The interactions of dipole and multipole moments, hydrogen bonds, and charged atoms in molecules are included in the hydrophilic interactions in this model. The HMLP is used to study hydrogen bonds and water-octanol partition coefficients in several examples. The calculated results show that HMLP gives qualitatively and quantitatively correct, as well as chemically reasonable results in cases where comparisons are available. These comparisons indicate that the HMLP has advantages over the empirical lipophilicity potential in many aspects. Three possible screening functions and parameters used in them are tested and optimized in this research. The power screening function, bi||Ri - r||ã, and the exponential screening function, biexp(-||Ri - r|| / d0), give satisfactory results. A new strategy for drug design and combinatory chemistry is presented based on HMLP, and is used in the study of a small molecular system, pyrazole and its derivatives. The mechanism of inhibition of LADH caused by pyrazole and its derivatives is explained based on the calculation results of HMLP indices. Good results are achieved in this example. Further improvements of screening function and visualization of HMLP by computer graphics are discussed. I suggest two possible visualization approaches of HMLP: a two-color system and a three-color system. Their possible applications are discussed. HMLP is suggested as a potential tool in computer-aided three-dimensional drug design, studies of 3D-QSAR, active structure of proteins, and other types of molecular interactions.