GAPE: An Improved Genetic Algorithm for Pharmacophore Elucidation

Prediction of the binding mode for a series of active compounds, in the absence of known protein structure, is a problem of paramount importance in rational drug design. GAPE (genetic algorithm for pharmacophore elucidation) is an automated multicompound overlay creation program, based on the original GASP program, that uses a genetic algorithm to fully explore the conformational space of the input structures and their alignments, so as to elucidate a pharmacophore. The software was evaluated on 13 test systems from nine protein targets using overlaid ligands extracted from the PDB. Using objective rmsd criteria and starting from 2D structures, in the absence of any protein information, GAPE was observed in eight systems to approximate the crystallographically observed binding mode. In the predicted alignments for each of those eight systems, at least half the input structures were within 2 Å rmsd of the crystal structure coordinates. Further analysis, using stricter subjective criteria, showed considerable success in five systems. For example, the prediction for a set of 12 ligands targeting P38 had 11 ligands with a 1.8 Å rmsd to crystal structure coordinates. Finally, the algorithm was favorably compared with the current GASP and Galahad programs.