有机污染物生物降解性预测模型

李彦莹; 李雪花; 杨先海; 陈广超; 陈景文

生态毒理学报

有机污染物生物降解性预测模型

, , , ,

李彦莹, 李雪花, 杨先海, 陈广超, 陈景文. 有机污染物生物降解性预测模型[J]. 生态毒理学报, 2012, 7(5): 549-555.

引用本文:

李彦莹, 李雪花, 杨先海, 陈广超, 陈景文. 有机污染物生物降解性预测模型[J]. 生态毒理学报, 2012, 7(5): 549-555.

Li Yanying, Li Xuehua, Yang Xianhai, Chen Guangchao, Chen Jingwen. Predicting Models for Biodegradability of Organic Pollutants[J]. Asian journal of ecotoxicology, 2012, 7(5): 549-555.

Citation:

Li Yanying, Li Xuehua, Yang Xianhai, Chen Guangchao, Chen Jingwen. Predicting Models for Biodegradability of Organic Pollutants[J]. Asian journal of ecotoxicology, 2012, 7(5): 549-555.

有机污染物生物降解性预测模型

大连理工大学环境学院工业生态与环境工程教育部重点实验室, 大连 116024

作者简介: 李彦莹(1989-),女,学士,研究方向:计算毒理学;E-mail:yyli@mail.dlut.edu.cn

基金项目:
国家863计划重点项目(2010AA065105)和工业生态与环境工程教育部重点实验室开放课题
中图分类号: X171.5

Predicting Models for Biodegradability of Organic Pollutants

Liaoning Key Laboratory of Industrial Ecology and Environmental Engineering, Ministry of Education, School of Environmental Science and Technology, Dalian University of Technology, Dalian 116024, China
Fund Project:

摘要: 生物降解性是评估污染物环境持久性的重要依据,也是化学品是否获准生产及进入市场的评价指标。采用17位生物降解领域专家评估的生物降解等级数据,通过功能树(FT)算法建立了包含15个分子结构参数的初级生物降解和最终生物降解预测模型。外部验证结果表明,模型具有较好的预测准确性,初级生物降解性加权准确度(weighted accuracy,WA):训练集WA=84.1%,验证集WA=78.9%;最终生物降解性WA:训练集WA=91.0%;验证集WA=83.6%。预测正确性对化合物的杠杆值作图,表征了生物降解性模型的应用域。
- 有机污染物 /
- 初级生物降解性 /
- 最终生物降解性 /
- 预测模型 /
- 功能树
Abstract: Biodegradability is significant for the evaluation of environmental persistence of pollutants, which has become a criterion of evaluation on a chemical allowing to be manufactured and market. The primary biodegradation and ultimate biodegradation models which contained 15 molecular structure descriptors were developed employing 17 biodegradation experts evaluating rating data and functional tree algorithm. The external validation results indicated that the models have a good predictive ability. The weighted accuracy (WA) of primary biodegradation is 84.1% and 78.9% for training set and validation set, respectively. WA for ultimate biodegradation is 91.0% and 83.6% for training set and validation set, respectively. Moreover, the applicability domain of the biodegradation models were characterized by the plots of prediction accuracy versus leverage values of chemicals.
- organic pollutants /
- primary biodegradability /
- ultimate biodegradability /
- prediction model /
- functional tree

点击查看大图

计量

文章访问数: 1744
HTML全文浏览数: 1744
PDF下载数: 57
施引文献: 0

李彦莹, 李雪花, 杨先海, 陈广超, 陈景文. 有机污染物生物降解性预测模型[J]. 生态毒理学报, 2012, 7(5): 549-555.

引用本文:

李彦莹, 李雪花, 杨先海, 陈广超, 陈景文. 有机污染物生物降解性预测模型[J]. 生态毒理学报, 2012, 7(5): 549-555.

Li Yanying, Li Xuehua, Yang Xianhai, Chen Guangchao, Chen Jingwen. Predicting Models for Biodegradability of Organic Pollutants[J]. Asian journal of ecotoxicology, 2012, 7(5): 549-555.

Citation:

Li Yanying, Li Xuehua, Yang Xianhai, Chen Guangchao, Chen Jingwen. Predicting Models for Biodegradability of Organic Pollutants[J]. Asian journal of ecotoxicology, 2012, 7(5): 549-555.

有机污染物生物降解性预测模型

作者简介: 李彦莹(1989-),女,学士,研究方向:计算毒理学;E-mail:yyli@mail.dlut.edu.cn
大连理工大学环境学院工业生态与环境工程教育部重点实验室, 大连 116024

收稿日期: 2012-06-29

基金项目:

国家863计划重点项目(2010AA065105)和工业生态与环境工程教育部重点实验室开放课题

关键词:

摘要: 生物降解性是评估污染物环境持久性的重要依据,也是化学品是否获准生产及进入市场的评价指标。采用17位生物降解领域专家评估的生物降解等级数据,通过功能树(FT)算法建立了包含15个分子结构参数的初级生物降解和最终生物降解预测模型。外部验证结果表明,模型具有较好的预测准确性,初级生物降解性加权准确度(weighted accuracy,WA):训练集WA=84.1%,验证集WA=78.9%;最终生物降解性WA:训练集WA=91.0%;验证集WA=83.6%。预测正确性对化合物的杠杆值作图,表征了生物降解性模型的应用域。

English Abstract

Predicting Models for Biodegradability of Organic Pollutants

Liaoning Key Laboratory of Industrial Ecology and Environmental Engineering, Ministry of Education, School of Environmental Science and Technology, Dalian University of Technology, Dalian 116024, China

Received Date: 2012-06-29

Fund Project:

Keywords:

Abstract: Biodegradability is significant for the evaluation of environmental persistence of pollutants, which has become a criterion of evaluation on a chemical allowing to be manufactured and market. The primary biodegradation and ultimate biodegradation models which contained 15 molecular structure descriptors were developed employing 17 biodegradation experts evaluating rating data and functional tree algorithm. The external validation results indicated that the models have a good predictive ability. The weighted accuracy (WA) of primary biodegradation is 84.1% and 78.9% for training set and validation set, respectively. WA for ultimate biodegradation is 91.0% and 83.6% for training set and validation set, respectively. Moreover, the applicability domain of the biodegradation models were characterized by the plots of prediction accuracy versus leverage values of chemicals.