Validation of SOMFA using Data Mining Technique

Abstract- In drug design, the investigation of properties of chemical compounds is the most important task. For determining the properties, the analysis of the existing data set is essential. Instead of describing individual molecules, in drug design, methods are used to characterize complete sets of chemical compounds and their relationship. Data mining analyzes large amount of data to obtain useful information leading to understanding of relationships within chemical compounds to extract "hidden " information for decision making. This paper describes the various data mining techniques used in Cheminformatics to analyze chemical data sets for molecular patterns, for extraction of relevant information and the production of reliable secondary information for drug discovery. Data mining techniques also helpful for construction of statistical model (QSAR Model) that is useful to testing of validation and reliability of Cheminformatics tools. Validation is a crucial element for design of any tool. The reliability of any tool/model depends on how well the tool can predict the activity of compounds outside the training set and reproduces the biological activity of compounds included in the model. In this paper we validate SOMFA tool by using QSAR model on dataset taken from literature