A QSTR Model For Toxicity Prediction Of Pesticides Towards Fleas

A large amount of residual pesticides in water bodies not only causes pollution to the water environment,but also affects the health of aquatic organisms.Daphnia magna is an ideal experimental species in the ecotoxicological evaluation of chemicals.Also,the U.S.Environmental Protection Agency listed the acute toxicity test of Daphnia magna as a mandatory item for new pesticide registration.Animal test to obtain toxicity data for a large number of pesticides is time-consuming,expensive and complicated,and violates the"3Rs-principle"of animals ethics.Thus,it is necessary to develop theoretical calculation methods to evaluate the acute toxicity of pesticides to Daphnia magna.Quantitative structure toxicity relationship（QSTR）methods have been widely used in the hazards assessment of chemicals and pharmaceuticals,and are effective tools to determine the relationship between the molecular structures and their biological activity/physicochemical properties.Based on the norm index concept proposed by our research group,combined with density functional theory（DFT）to optimize the structure of pesticides,a new set of norm descriptors was constructed.Using the multiple linear regression method,a generalized QSTR model based on a large data sample set was developed and successfully applied to the prediction of the acute toxicity of pesticides to Daphnia magna.The main findings are as follows:A model for predicting the acute toxicity(48h-p EC₅₀)of various pesticides to Daphnia magna was established.The model statistical parameters R_training²=0.8045 and R_testing²=0.8224 indicate that the model has good predictive ability.Internal validation and external validation show the accuracy and stability of the developed model.The Y-randomization test shows that there is no random correlation and over fitting in the modeling process.The application domain analysis result shows that the working model has a wide application range.Compared with other literature research work,the result shows that the model has wider application scope and more accurate prediction.Meantime,the acute toxicity of 67 pesticide molecules without experimental toxicity data to Daphnia magna was predicted by this working model,and some promoters affecting the toxicity of pesticides were clearly pointed out,which provided a reference for the experimental determination of the toxicity of these pesticide molecules.The method of density functional theory（DFT）parameters combined with norm descriptors proposed in this study is effective for the QSTR model,which can accurately describe the relationship between the molecular structure and biological toxicity of various pesticides.The accurate and reliable QSTR model established in this work can provide a scientific basis for the molecular design of new pesticides and the evaluation of ecological risk toxicity.