Study On Structure-toxicity Relationship Of Organic Compounds To Fish

Organic compounds play an essential role in people's daily life and production in industry and agriculture.The aquatic ecosystem and even human health will be potentially threatened by organic compounds released into the environment in a variety of ways.It is necessary to evaluate the potential environmental risk of organic compounds in order to prevent environmental pollution and regulate the effective supervision of organic compounds.The biological toxicity information of organic compounds to fish is an important fundamental data for ecological risk assessment.However,there are more than one hundred billion organic compounds in existence,and it is time-consuming,labor-consuming and involves ethical issues to obtain toxicity data by the experimental method.Quantitative structure-activity relationship(QSAR)is thus an effective alternative to obtain toxicity information.Based on the concept of norm descriptors proposed by our group,a series of atom distribution matrices were constructed to obtain norm descriptors,and several QSAR models were established by using the method of multiple linear regression.Meanwhile,in order to analyze the stability and the prediction ability of these models,cross validation,internal validation,Y-randomized analysis and applicability domain analysis were utilized to verify these models.The main contents of this work are as follows:Study on the QSAR for acute toxicity of pesticide organic compounds toward rainbow trout was performed.In the present work,two norm expressions were proposed and several norm descriptors were derived to establish a QSAR model for predicting the acute toxicity of various pesticide organic compounds to rainbow trout.The acute toxicity of 311 pesticide organic compounds to rainbow trout was calculated.The overall squared correlation coefficient R2 of the established QSAR model was 0.8053,the result of squared correlation coefficient of the leave-one-out cross-validation was QLOO2 with 0.7606,and R2 of the training set and testing set were Rtraining2=0.8011 and Rtesting2=0.8108,respectively.The results of validation and comparison of references revealed that the QSAR model established by our method was effective and reliable,and the sample size and accuracy of the present work were superior to those of literature works.Study on the QSAR for acute toxicity of organic compounds toward zebrafish embryos was carried out.In this study,the norm descriptor was obtained by calculating the norm index of the atomic distribution matrices,which were defined according to atomic spatial distribution and atomic properties.Four QSAR models were established to describe the acute toxicity of various organic compounds to zebrafish embryos at different exposure times.The squared correlation coefficient R2 of these norm index-based QSAR models at different exposure times of 48 hours,96 hours,120 hours and 132 hours was 0.8549,0.9162,0.8335 and 0.8119,respectively.The results of QLOO2 for the leave-one-out cross validation were 0.8140,0.8786,0.7732 and 0.7022,respectively.Statistical results and validation results indicated that the four QSAR models corresponding to different exposure times established in the present work were stable and reliable,and had a wide range of applicability domain.Therefore,these norm descriptors may be effective in predicting the acute toxicity of various organics to zebrafish embryos.These results will also contribute to assess the potential risk of organic compounds to the aquatic environment.In the present study,QSAR models of the acute toxicity of organics to two kinds of fish was established based on the norm descriptor,and the accuracy,stability and reliability of the model were verified by evaluation of model validation and literature comparison.The QSAR models developed in the present work could describe the acute toxicity of different kinds of organic compounds to rainbow trout and zebrafish embryo,indicating the effectiveness of the proposed norm descriptor,which is helpful for the environmental risk assessment of organic compounds.The norm descriptors proposed by our group are directly obtained from the molecular structure,and without relying on other macroscopic basic property data.Thus,norm descriptors are the comprehensive essential reflection of the molecular structure,which might be with the accuracy,generalization and universality.