| Popis: |
Database searching plays an increasingly important role in drug-discovery programs [1]. Facilities for substructure searching, the identification of all of those molecules in a database that contain a user-defined query substructure, have been available in chemical information systems for many years [2]. The last few years have seen the introduction of complementary facilities for similarity searching [3]. This involves matching a target molecule of interest, such as a weak lead from a high-throughput screening program, against all of the molecules in a database to find the nearest neighbors — i.e. those molecules that are most similar to the target using some quantitative measure of intermolecular similarity. Early database searching systems were designed for the storage and retrieval of twodimensional (2D) chemical structures but the development of structure generation programs [4] has focused interest on techniques for the processing of three-dimensional (3D) structural information [5], and there have already been several reports of systems for 3D similarity searching that are sufficiently fast in operation to allow them to be used with databases of non-trivial size [6–12]. However, few of these approaches take explicit account of the electrostatic, steric and hydrophobic fields that form the basis of modern approaches to 3D QSAR (as illustrated by the many other papers in this volume), and an ongoing project at the University of Sheffield is hence developing methods for field-based similarity searching. Like many previous workers [13–21], our experiments have focused on the Molecular Electrostatic Potential (MEP), but the techniques that we have developed are applicable, in principle at least, to any field-like attribute that can be represented by real values in a 3D grid surrounding a molecule. The electrostatic potential, Pr, at a point r for a molecule of n atoms is calculated from the point charges qi on each atom i in the molecule, so that |
