Molecular modeling: Application of Support Vector Machines and Decision trees for anti-HIV activity prediction of organic compounds

Abstract

Multivariate methods of pattern recognition, classification and discriminant analysis have been found most useful in many types of chemical and biological problems. Predicting the biological activity of molecules from their chemical structures is a principal problem in drug discovery. Pattern recognition has gained attention as methods covering this need. In the present study classification models for inhibiting Human Immunodeficiency Virus (HIV) activity, based on Support Vector Machines (SVM) and Decision trees (DT), are developed. The obtained results indicate that SVM and DT can be employed as a forceful tool for quantitative structure-activity relationship studies.