APPLICATION OF NEURAL NETWORKS TO THE STUDY OF STRUCTURE-ACTIVITY-RELATIONSHIPS OF 6,7-DIARYLPTERIDINES AS NEMATOCIDES

A study of structure-activity relationships of 6,7-diarylpteridines as nematocide agents using a trained back-propagation neural network has been carried out. The use of this network has allowed the prediction of the qualitative nematocide activity of pteridine derivatives not ye t synthesized. Among twenty-five pteridine derivatives, presclected, s eventeen of them were predicted as nematocides by the network. The syn thesis and the nematocide activity of the pteridines, which had been p redicted as active compounds, are reported. The results obtained have shown that the use of this network allows the prediction of qualitativ ely nematocide activity of pteridine derivatives, not yet synthesized, in a high percentage.