A Hybrid Approach Based on Self-Organizing Neural Networks and the K-Nearest Neighbors Method to Study Molecular Similarity

The “Molecular Similarity Principle” states that structurally similar molecules tend to have similar properties—physicochemical and biological. The question then is how to define “structural similarity” algorithmically and confirm its usefulness. Within this framework, research by similarity is registered, which is a practical approach to identify molecule candidates (to become drugs or medicines) from databases or virtual chemical libraries by comparing the compounds two by two. Many statistical models and learning tools have been developed to correlate the molecules’ structure with their chemical, physical or biological properties. The role of data mining in chemistry is to evaluate “hidden” information in a set of chemical data. Each molecule is represented by a vector of great dimension (using molecular descriptors), the applying a learning algorithm on these vectors. In this paper, the authors study the molecular similarity using a hybrid approach based on Self-Organizing Neural Networks and Knn Method. process of laboratory experimentation. This process, from hit to lead to marketable drug, is typically as long as 5-10 years. In order to identify new molecules susceptible to become medicines, the pharmaceutical research has more and more resort to technologies permitting DOI: 10.4018/ijcce.2011010106 76 International Journal of Chemoinformatics and Chemical Engineering, 1(1), 75-95, January-March 2011 Copyright © 2011, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. to synthesize a very big number of molecules simultaneously and to test their actions on a given therapeutic target. These data can be exploited to construct the models permitting to predict the properties of molecules not yet tested, even not yet synthesized. Looking for molecular similarity is an intelligent way to design drug. Its use is based on the principle that structurally more similar molecules are more likely to exhibit similar properties than structurally less similar molecules (Monev, 2004; Johnson & Maggiora, 1990). Such predictive models are very important because they make it possible to suggest the synthesis of new molecules, and to eliminate very early in the molecule’s search process the molecules whose properties would prevent their use as medicine. We speak then of virtual sifting. Hence, searching for functionally similar molecules, which is very important in drug design, can be accomplished by searching for structurally similar molecules (van de Waterbeemd & Gifford, 2003). But the problem is to define molecular similarity.