Study of modeling and optimization for predicting the acute toxicity of carbamate pesticides using the binding information with carrier protein

Abstract

To predict drug acute toxicity using the binding information with human serum albumin, our research group established a new method (Carrier protein binding information-toxicity relationship, CPBITR). Unfortunately, the previous model had too few data sets which may affect the accuracy and credibility of the model. In this paper, therefore, we measured the binding modes of three carbamate pesticides, Bendiocarb, Butocarboxim and Dioxacarb with human serum albumin (HSA) to supplement the previously modeled training set. Multispectral methods and molecular docking were used to study their binding modes. We built and optimized the previous models with the combined information of three different toxicity pesticides and HSA in order to find better prediction method. The results showed that Back-propagation Artificial Neural Network model has the best fitting effect among these models. In conclusion, the proposed model effectively improves the accuracy and credibility of the existing model. It results in significant predict drug acute toxicity using the binding information with carrier protein and contribute to drug development and research.