Artificial neural network model based QSAR for oxygen containing heterocycles as selective COX-2 inhibitors

Quantitative structure activity relation (QSAR) towards COX-2 inhibitoryactivity were developed for three oxygen containing heterocycles namely, 3,4,6-triphenylpyran-2-ones, 2,3-diarylpyran-4-ones and 5-aryl-2,2-dialkyl-4-phenyl-3(2H)furanones. Regression and artificial back propagation neural network models were testedfor fitting the data. For the individual data sets octanol-water partition coefficient,geometric and connectivity indices were highly correlated with activity. For the combineddata set the extent of branching, and molecular shape factor had a positive correlation andmolecular connectivity index had a negative correlation with COX-2 inhibitory activity. A4-2-1 back propagation neural network model fitted the combined data set well (R2= 0.77,R2adj =0.73, q2 = 0.63, and F=243). The predictive capability of the neural network modelwas gauged by using part of the furanone data for learning and the rest for validation andsystematically reducing the number of processing elements in the hidden layer. © 2012 Bentham Science Publishers. All rights reserved.