EXPERIMENTAL SYSTEM FOR SIMILARITY AND 3D SEARCHING OF CAS REGISTRY SUBSTANCES .1. 3D SUBSTRUCTURE SEARCHING

Chemical Abstracts Service (CAS) is developing an experimental system for similarity and 3D searching on CAS Registry substances. The purpose of this system is to obtain user input on desirable capabilities and data content for such searching. Currently, the system supports 3D exact, substructure, and superstructure searching. The 3D coordinates for the system's databases were generated via the CONCORD program. These databases include a CAS 3D structure templates (CAST-3D) subset of over 365 000 substances with limited conformational flexibility. The experimental system utilizes a client-server architecture using client workstations for query framing and display and a single search engine compute server. Search levels include a screen step followed by atom-by-atom search (using a modified Ullman subgraph isomorphism algorithm), and, where appropriate, a geometric superimposition of the query and answer file substance. Two logical 3D substructure query types are supported: a general query, typically used for pharmacophore pattern matching, and a framework query, typically used in locating synthetic precursors that lead to a desired geometric orientation of substituents. Novel screens based on atom triangle and tetrangle distances as well as global and localized flexibility indices provide for effective and efficient screening. Also, user parameters can specify approximate local and global conformational flexibility characteristics for the matching file substances. This paper describes the features and capabilities that are currently available in the experimental system along with an illustrative application scenario.