A quantitative structure-activity relationship model for radical scavenging activity of flavonoids

A quantitative structure-activity relationship (QSAR) study has been carried out for a training set of 29 flavonoids to correlate and predict the 1,1-diphenyl-2-picrylhydrazyl radical scavenging activity (RSA) values obtained from published data. Genetic algorithm and multiple linear regression were employed to select the descriptors and to generate the best prediction model that relates the structural features to the RSA activities using (1) three-dimensional (3D) Dragon (TALETE srl, Milan, Italy) descriptors and (2) semi-empirical descriptor calculations. The predictivity of the models was estimated by cross-validation with the leave-one-out method. The result showed that a significant improvement of the statistical indices was obtained by deleting outliers. Based on the data for the compounds used in this study, our results suggest a QSAR model of RSA that is based on the following descriptors: 3D-Morse, WHIM, and GETAWAY. Therefore, satisfactory relationships between RSA and the semi-empirical descriptors were found, demonstrating that the energy of the highest occupied molecular orbital, total energy, and energy of heat of formation contributed more significantly than all other descriptors.