Study on the Balance Activity of Mice Based on BP-ANN

Abstract

Abstract In order to study the quantitative structure-activity relationship (QSAR) of the balance activity (pH) for 30 benzodiazepinooxazole derivatives to male mice, the molecular electronegativity distance vector (M D) of above compounds was calculated by program according to molecular topological environment in this paper. The five-variable (M 6, M 46, M 25, M 63, M 70) QSAR model of pH for above compounds was constructed by using leaps-and-bounds regression method. The result demonstrates that the model is robustness and good prediction ability by using R cv 2, F tests. The M 6, M 46, M 25, M 63 and M 70 were used as the input neurons of artificial neural network (ANN), and a 5:3:1 network architecture was employed. A satisfying BP-pH model could be constructed with the back-propagation algorithm, with the correlation coefficient (R 2) and the standard error(S D) being 0.928 and 0.117, respectively, showing that the relationship between pH and these structural parameters has a good nonlinear correlation. Form the three parameters of the model, it is known that the dominant influence factors of increased balance activity are the microscopic fragments: –CH3, –CH2–, >C<, –NH–, –N< and –X in the molecules.