Medicinal and Natural Products Chemistry Research Center, Shiraz University of Medical Sciences, Shiraz, Iran.
Ecotoxicol Environ Saf. 2013 Jan;87:42-8. doi: 10.1016/j.ecoenv.2012.10.005. Epub 2012 Oct 27.
Many ionic liquids are soluble in water and their impact on the aquatic environment has to be evaluated. However, due to the large number of ionic liquids and lack of experimental data, it is necessary to develop estimation procedures in order to reduce the materials and time consumption. In this study using multilayer perceptron neural network (MLP), ant colony optimization (ACO) and multiple linear regression (MLR) strategies, good predictive quantitative structure-activity relationships (QSAR) models were introduced and structural parameters affecting ecotoxicity of ionic liquids in limnic green algae (Scenedesmus vacuolatus) were revealed. Moreover, principal component analysis (PCA) and cluster analysis (CA) approaches were also applied to visualize any possible patterns or relationships among ionic liquids data. It was revealed that selected descriptors of the MLR model are also capable of clustering ionic liquids according to their four level of toxicity.
许多离子液体可溶于水，因此必须评估其对水生环境的影响。然而，由于离子液体数量众多且缺乏实验数据，因此有必要开发估算程序以减少材料和时间的消耗。在本研究中，使用多层感知器神经网络（MLP）、蚁群优化（ACO）和多元线性回归（MLR）策略，引入了良好的预测定量构效关系（QSAR）模型，并揭示了影响离子液体在湖泊绿藻（Scenedesmus vacuolatus）中生态毒性的结构参数。此外，还应用主成分分析（PCA）和聚类分析（CA）方法直观地显示离子液体数据之间的任何可能的模式或关系。结果表明，MLR 模型选择的描述符也能够根据离子液体的四级毒性对其进行聚类。
