PHARMACOPHORIC PATTERN-MATCHING IN FILES OF 3-DIMENSIONAL CHEMICAL STRUCTURES - IMPLEMENTATION OF FLEXIBLE SEARCHING

The conformational space of a flexible three-dimensional (3D) molecule can be represented for searching purposes by a smoothed bounded distance matrix. Such matrices provide an effective way of carrying out flexible searching, but search times can be very long when compared with rigid searches that take no account of conformational flexibility. This paper considers two techniques for minimizing the computational requirements of flexible searching. In the first part, we compare four different indices that have been suggested for the quantification of molecular flexibility, and demonstrate that they produce comparable rankings of sets of molecules in order of decreasing flexibility. We then demonstrate that the prioritization of a set of structures in order of increasing flexibility can result in substantial reductions in the time requirements of flexible searching if some fixed number of hit structures is desired. In the second part, we report an analysis of the 3D crystal structures in the Cambridge Structural Database to generate distance ranges that are tighter than those produced by distance geometry. Experiments with a set of six query pharmacophores demonstrate that use of these tightened ranges results in substantial reductions in search times, but that this may also lead to a reduction in the number of hit molecules obtained from the final conformational search.