MOLECULAR SUBSTRUCTURE SIMILARITY SEARCHING - EFFICIENT RETRIEVAL IN 2-DIMENSIONAL STRUCTURE DATABASES

Molecular substructure searching and full-structure similarity searching have become standard structure retrieval techniques in chemical database systems. While these search methods often provide useful results, they do not always meet the needs of the chemical database searcher. This paper describes the development of a substructure similarity (subsimilarity) search system that finds molecules containing a substructure that is similar to a query structure. It is argued that subsimilarity searching fulfills a retrieval need not met by existing search techniques. A major goal in the design of the system, the ability to execute a majority of subsimilarity searches interactively, was realized through a two-step search method that employs a unique subsimilarity screening step followed by an approximate maximal common subgraph (MCS) matching step. User acceptance of subsimilarity searching and its application in the research environment are discussed within the context of Upjohn's Cousin chemical database system.