Exploiting Molecular Diversity: Pharmacophore Searching and Compound Clustering
Vincent J. van Geerestein, Hans Hamersma, Steven P. van Helden · 1997
Pharmacophore searching and compound clustering are important concepts of computer based methods for lead finding strategies. These concepts are very much linked to modern developments in drug discovery and they are the direct response of computer chemists to the current achievements of combinatorial chemistry (CC) and high throughput screening (HTS). Pharmacophore searching is now an established technique which has a proven track record in many discovery programs of pharmaceutical companies. The outstanding issue here is probably the problem of dealing with conformational flexibility. In the paper we will share our experiences using conformationally flexible pharmacophore searching and give an outline of the strategy used in our group. In the hit list of a pharmacophore search, we typically obtain a factor of 10 of enrichment of the biological activity of concern. Compound clustering is now regularly used for rational selection of compounds for acquisition for in-house compound collections used for screening. However, also here challenges and dilemmas remain. Clustering is typically done on the basis on 2D fingerprints, derived from techniques used for fast substructure searching by scientists working in the field of chemical information. We show that comparable results may be obtained with 3D fingerprint derived from 3D pharmacophores. Compound clustering in lead finding appears to be of value as a method of prioritising the test order or in case the throughput of the assay prevents testing all compounds. In the same way it may be used to select and prioritise compounds for acquisition.