有机化合物结构与生物降解性分类模型研究  

Study on Chemical Structure of Organic Compounds and Biodegradation Classification by SVC Model

作　　者：胡俊杰[1] 周红[1] 周林军[2] 

机构地区：[1]环境保护部化学品登记中心,北京100012 [2]环境保护部南京环境科学研究所,江苏南京210042

出　　处：《环境科学与技术》2014年第9期55-58,共4页Environmental Science & Technology

基　　金："863"计划课题:化学品种类快速甄别及毒性模型测试技术(2012AA06A301);"863"计划课题:化学品暴露和效应评估及检测关键技术(2013AA06A308)

摘　　要：采集包含我国近5 a申报的新化学物质在内共172种有机化学物质的28 d快速生物降解性实验测试数据,运用支持向量分类方法建立生物降解性预测分类模型,并应用该模型对测试集化合物和10种验证测试物质进行预测。预测结果表明,该模型具有很好的预测分类能力,有效反映了有机化合物结构与生物降解性间的内在关系,可广泛用于化学物质生物降解性的初步估算,确定进一步的化学品管理与实验测试要求。Data of fast aerobic biodegradation rate of 172 organic compounds in 28 days, among others, 84 new chemical substances from the information issued in the new chemical notification in China were collected, and the support vector machine was used to establish the predictive and classification model, SVC model, for the organic compounds. To verify the prediction and classification ability, the self-built model was applied to the prediction of 43 compounds in the test set and 10 experimental compounds. The results indicated that the self-built model had a good classification ability, which made it possible to reveal the intrinsic relationship between the structure and biodegradation. The model could be widely used in estimation of biodegradation for compounds, especially the new chemical substances.

关 键 词：快速生物降解性 新化学物质 支持向量分类模型 

分 类 号：X32[环境科学与工程—环境工程]