Mining a large database for peptidomimetic ring structures using a topological index

A topological index that characterizes the way in which a substructure is imbedded in the rings of a structure was recently introduced as a tool for analyzing large substructure-search answer sets. This paper describes a technique that uses the ring-imbedding index to mine a chemical database for novel compounds. The strategy is to run a very general substructure search and then extract from the answer set those structures with the least common index values. An experimental test was performed on the CAS Registry File. Using a peptide-like substructure query that retrieved over 33 000 answers, this technique was able to detect a number of unusual peptidomimetic ring structures in the database. This technique is potentially useful when the problem of incorporating rings into a particular type of structure is of interest, as is often the case in medicinal chemistry.