Classification scheme for the design of serine protease targeted compound libraries
Authors: Lang, S.A.; Kozyukov, A.V.; Balakin, K.V.; Skorenko, A.V.; Ivashchenko, A.A.; Savchuk, N.P.
Source: Journal of Computer-Aided Molecular Design, Volume 16, Number 11, November 2002 , pp. 803-807(5)
The development of a scoring scheme for the classification of molecules into serine protease (SP) actives and inactives is described. The method employed a set of pre-selected descriptors for encoding the molecular structures, and a trained neural network for classifying the molecules. The molecular requirements were profiled and validated by using available databases of SP- and non-SP-active agents [1,439 diverse SP-active molecules, and 5,131 diverse non-SP-active molecules from the Ensemble Database (Prous Science, 2002)] and Sensitivity Analysis. The method enables an efficient qualification or disqualification of a molecule as a potential serine protease ligand. It represents a useful tool for constraining the size of virtual libraries that will help accelerate the development of new serine protease active drugs.
Document Type: Research Article
Affiliations: Chemical Diversity Labs, Inc., 11558 Sorrento Valley Road, San Diego, CA 92121, USA
Publication date: November 1, 2002