Chemometrical treatment of electronic structures of 28 flavonoid derivatives
Marjan Vra ko, Marjana Novi?, Marko Perdih · International Journal of Quantum Chemistry · 2000
Electronic structures of 28 flavonoid derivatives were treated with the chemometrical methods, i.e., multiple linear regression, counterpropagation artificial neural networks, principal component analysis, and partial least-squares method. Electronic structures given as density-of-states spectra were constructed from calculated molecular orbital energies. It was shown that all the methods recognize some structural features from electronic structures. The relationship between electronic structures and biological activity was also studied. The reported results, i.e., the predicted activities, are comparable to the results obtained by models built up with standard descriptors, i.e., topological indices, geometric parameters, and electrostatic indices or “spectrum-like representations of three-dimensional structures.” For linear models the correlation coefficients obtained by a cross validation leave-one-out method are close to 0.9, and for artificial neural network models they are about 0.8. © 2000 John Wiley & Sons, Inc. Int J Quant Chem 76: 733–743, 2000