Loading...

Inspection of Inhibitory Effect of 5-Hydroxy-3(2H)-Pyridazinone Derivatives on Hepatitis C Virus Using Chemometric Methods

Paeenmahali, Habibollah | 2010

1527 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 40645 (03)
  4. University: Sharif University of Technology
  5. Department: Chemistry
  6. Advisor(s): Jalali-Heravi, Mehdi
  7. Abstract:
  8. The derivatives of 5-Hydroxy-3(2H)-pyridazinone show inhibitory effects on hepatitis C virus. The aim of the present work was modeling and prediction of inhibitory effects of these derivatives (Log(1/EC50)) on this disease. In this research, a data set of 119 molecules of 5-Hydroxy-3(2H)-pyridazinone derivatives that have inhibitory effect on hepatitis C virus was selected. The MLR model was generated using SPSS package. Five important descriptors were selected applying stepwise variable selection technique. These descriptors selected through 1207 descriptors that were calculated for all molecules in data set. Best model with high R2 and F values and low RMSE was selected for the prediction of the experimental values. Results of the MLR model were not satisfactory. To assess the nonlinear characteristics of inhibitory effect artificial neural networks (ANN) were used. In the ANN modeling the number of nodes in the hidden layer, momentum, learning rate, number of iterations and biases were optimized. To reduce the error of prediction, genetic algorithm (GA) was employed to select the most important variables. The ANN was used to construct a model between the selected variables and EC50. Comparison of MLR, MLR-ANN and GA-ANN shows that GA-ANN is superior over the others from point of the prediction ability. Inspection of the selected descriptors reveals that electronic and geometric parameters are the most important ones affecting the activity of these derivatives on hepatitis C virus.
  9. Keywords:
  10. Neural Network ; Genetic Algorithm ; Multiple Linear Regression Analysis ; Hepatitis C Virus ; 5-Hydroxy-3 (2H)Pyridazinone

 Digital Object List

 Bookmark