明亮发光杆菌
硒腹
数量结构-活动关系
偏最小二乘回归
生化工程
生物系统
水生毒理学
有机化学品
分子描述符
化学毒性
毒性
环境化学
化学
计算机科学
生物
机器学习
有机化学
工程类
立体化学
作者
Mainak Chatterjee,Kunal Roy
标识
DOI:10.1016/j.jhazmat.2020.124936
摘要
The rapid industrialization has led to the generation of various organic chemicals and multi-component mixtures which affect the environment adversely. Although organic chemicals are often exposed to the environment as a form of chemical mixtures rather than individual compounds, there is insufficient toxicity data available for the chemical mixtures due to the associated complexities. Most importantly, the nature of toxicity of mixtures is completely different from the individual chemicals, which makes the evaluation more difficult and challenging. In this paper, we have developed QSAR models for various individual and mixture data sets for the prediction of the aquatic toxicity. We have used Partial Least Squares (PLS) regression as a statistical tool to build the models. The various structural features of the individual chemicals and the mixture components have been modeled against the toxicity end point pEC50 (negative logarithm of median effective concentration in molar scale) of the aquatic organisms Photobacterium phosphoreum (marine bacterium) and Selenastrum capricornutum (freshwater algae). The mixture descriptors have been calculated by the weighted descriptor generation approach. The models were developed in accordance with OECD guidelines, and the quality of each model has been adjudged by strict validation parameters. The final models are robust, extremely predictive and interpretable mechanistically which can be used for the prediction of toxicity of untested chemical mixtures under the domain of applicability of the developed models.
科研通智能强力驱动
Strongly Powered by AbleSci AI