The quantitative structure - activity relationship of novel cephalosporin derivatives
Hao Li-fen, Zhou Lu, Gang Li · Huaxi yaoxue zazhi · 2005
OBJECTIVE To study the QSAR on the antibacterial activities of a series of novel orally active cephalosporin derivatives against Haemophilus influenzae. METHODS The method of quantum chemistry and Back-propagation network were used. RESULTS 18 molecular descriptors of the 48 cephalosporin compounds have been calculated and 7 of them were selected. The QSAR model has been built with selected 6 molecule descriptors as inputs and the minimal inhibition concentration of cephalosporins to Haemophilus influenzae as outputs. 43 compounds were selected stochastically as the training set and while the residual 5 compounds as the prediction set for the neural network. The predicted values were in good agreement with the experimental values (the maxium relative error of prediction set 5%). CONCLUSION LUMO energy,LogP and the net charges of both the substituents of C 3 and C 8 have great effect on the activities of cephalosporin derivatives. Based on the calculation results, the neural network can effectively analyze the relationship of molecular descriptors and the inhibition activities of cephalosporin compounds.