Study On The Relationship Between Organic Structure And Logk_ow,the Eco-Toxicity

N-octanol-water partition coefficient is widely used in the environmental,agricultural and pharmaceutical industries for the risk and the activity assessment of organic chemicals.Benzene derivatives are typical pollutants in water environment,and their eco-toxicity risks are of great concern.Obtaining the physicochemical property or aquatic toxicity data of organic compounds by experimental methods is not only time consuming but also costly.Quantitative structure-activity/property relationship(QSAR/QSPR)is an alternative approach to solving such problems,which plays an important role in the study of the physicochemical property of organic matter or environmental risk assessment.Based on the atomic distribution matrixes of organics,two QSAR/QSPR models were established by using the norm index descriptors.Norm index descriptors were used in this topic.Based on the method proposed by our group,the following two aspects were studied.The main findings were:(1)Established a multiple linear regression(MLR)model to describe the physicochemical properties of 1021 organic compounds-the octanol-water partition coefficient.The statistical parameters obtained by the model were:the squared correlation coefficient(R2)of the overall regression was 0.9307,and R2 of the training set,the testing set and the verification set were 0.9302,0.9331 and 0.9242,respectively.The squared value of the correlation coefficient obtained by the leave-one-cross validation method was 0.9260.F value was 610.The prediction and verification results shown that the model established by this method was effective and reliable,and it was better than the previous work in terms of accuracy and error.(2)Description the eco-toxicity(pEC so)of 110 benzene derivatives against V.fischeri.In this part of the work,using the norm descriptors,the QSAR model was established,and the eco-toxicity of V.fischeri was predicted for five benzene derivatives(aniline,phenol,thiophenol,benzoic acid and antibiotics).The training set and testing set were obtained.The R2 were 0.873 and 0.938,respectively.The main research conclusions were as follows:(?)The working model had good prediction accuracy for the eco-toxicity to V.fischeri;(?)Statistical results and model verification results indicated the stability and wide applicability of this model;(?)The norm descriptors constructed based on the atomic distribution matrix could accurately describe the molecular structure of organics,and the model established based on it was of great significance for the ecological risk assessment of organic chemicals.Based on the norm descriptors proposed by our group,this paper used QSAR/QSPR method to model and predict the physicochemical properties or eco-toxicity of organic pollutants.Through the internal verification,external verification,Y random verification,application domain analysis and comparison with other literature work,the accuracy,robustness and reliability of the model were verified.The results shown that the established model could better describe the octanol-water partition coefficients of organic compounds and the toxicity pEC50 of benzene derivatives against V.fischeri.And both showed that the descriptors proposed in this work were effective.The norm index descriptors of this group were obtained directly from the molecular structure,without relying on other basic data.It could be seen that the norm descriptors have accuracy,generality and universality.