USE OF STRUCTURE - ACTIVITY DATA TO COMPARE STRUCTURE-BASED CLUSTERING METHODS AND DESCRIPTORS FOR USE IN COMPOUND SELECTION

An evaluation of a variety of structure-based clustering methods for u se in compound selection is presented. The use of MACCS, Unity and Day light 2D descriptors; Unity 3D rigid and flexible descriptors and two in-house 3D descriptors based on potential pharmacophore points, are c onsidered. The use of Ward's and group-average hierarchical agglomerat ive, Guenoche hierarchical divisive, and Jarvis-Patrick nonhierarchica l clustering methods are compared. The results suggest that 2D descrip tors and hierarchical clustering methods are best at separating biolog ically active molecules from inactives, a prerequisite for a good comp ound selection method. Tn particular, the combination of MACCS descrip tors and Ward's clustering was optimal.