Chemometrics Modeling of MonoAmine Oxidase Inhibitory Effects of Pyrazoline Derivatives Using PCA-MLR-ANN Approaches

Please enable javascript in your browser.

Hassanzadeh Rad, Arman | 2010

665 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 40715 (03)
University: Sharif University of Technology
Department: Chemistry
Advisor(s): Jalali-Heravi, Mehdi
Abstract:
Prevalence of Multiple Drug Resistant Tuberculosis and use of Yersinia-Pestis in bioterrorism warrants synthesis of new antimicrobial agents. Although Pyrazoline derivatives were first synthesized as antimicrobial drugs, but they also had Mono Amine Oxidase inhibitory effects. In this Quantitative Structure – Activity Relationship (QSAR) study, MAO-I activity of pyrazoline derivatives were evaluated. By applying semi-empirical quantum calculations at AM1 level, optimum 3D geometry of 32 molecules were obtained.After descriptor generation, Principal Component Analysis (PCA) was performed to fiind out 5 outliers. After omitting outliers and sorting molecules according to IC50 values, Stepwise Multiple Linear Regression (MLR) was applied which lead to a multi-parametric equation contatining 4 descriptors and acceptable regression coefficients. Besides, Artificial Neural Network (ANN) with 4-4-1 architecture and MLR descriptors as inputs resulted in better regression coefficients. Both models show that spatial descriptors are significant in interaction of MAO with its inhibitors.
Keywords:
Principal Component Analysis (PCA) ; Artificial Neural Network ; Multiple Linear Regression Analysis ; Pyrazoline Derivatives ; MonoAmine Oxidase Inhibitors

No TOC