Using Multiple Linear Regression and Artificial Neural Network Techniques for Predicting CCR5 Binding Affinity of Substituted 1-(3, 3-Diphenylpropyl)-Piperidinyl Amides and Ureas

R. Mouhibi,M. Zahouily,Khalid El Akri,N. Hanafi

DOI: https://doi.org/10.4236/OJMC.2013.31002

2013-03-29

Abstract:Quantitative structure–activity relationship (QSAR) models were developed to predict for CCR5 binding affinity of substituted 1-(3, 3-diphenylpropyl)-piperidinyl amides and ureas using multiple linear regression (MLR) and artificial neural network (ANN) techniques. A model with four descriptors, including Hydrogen-bonding donors HBD(R7), the partition coefficient between n-octanol and water logP and logP(R1) and Molecular weight MW(R7), showed good statistics both in the regression and artificial neural network with a configuration of (4-3-1) by using Bayesian and Leven-berg-Marquardt Methods. Comparison of the descriptor’s contribution obtained in MLR and ANN analysis shows that the contribution of some of the descriptors to activity may be non-linear.