Kusić Hrvoje, Rasulev Bakhtiyor, Leszczynska Danuta, Leszczynski Jerzy, Koprivanac Natalija
Civil and Environmental Engineering Department, Jackson State University, 1400 J.R. Lynch Street, Jackson, Mississippi 39217, USA.
Chemosphere. 2009 May;75(8):1128-34. doi: 10.1016/j.chemosphere.2009.01.019. Epub 2009 Feb 7.
We present the results of the QSAR/QSPR study on the degradation rate constants of 78 aromatic compounds by the hydroxyl radicals in water. A genetic algorithm and multiple regression analysis were applied to select the descriptors and to generate the correlation models. Additionally to DRAGON descriptors, the parameters from quantum-chemical calculations at semiempirical and at density functional theory level (B3LYP/6-31G(d,p)) were applied. The most predictive model is a four-variable model that had a good ratio of the number of variables and the predictive ability to avoid overfitting. As it was expected, the main contribution to the degradation rate was given by the E(HOMO) parameter. Additionally, a number of topological descriptors in selected models showed an importance of polarizability term regarding the degradation rate of compounds. Overall, the applied GA-MLRA approach with the use of quantum-chemical and DRAGON generated descriptors showed good results in this study. The obtained statistically robust structure-degradation rate model can be used for future studies of the presence of organic compounds in the environment, and especially their degradation by hydroxyl radicals as a part of a water/wastewater treatment.
我们展示了关于78种芳香族化合物在水中被羟基自由基降解速率常数的定量构效关系/定量结构-性质关系（QSAR/QSPR）研究结果。应用遗传算法和多元回归分析来选择描述符并生成相关模型。除了Dragon描述符外，还应用了半经验和密度泛函理论水平（B3LYP/6-31G(d,p)）的量子化学计算参数。最具预测性的模型是一个四变量模型，该模型在变量数量与预测能力之间具有良好的比例，可避免过度拟合。正如预期的那样，E(HOMO)参数对降解速率的贡献最大。此外，所选模型中的一些拓扑描述符表明极化率项对化合物降解速率具有重要性。总体而言，在本研究中应用的使用量子化学和Dragon生成描述符的遗传算法-多元线性回归分析（GA-MLRA）方法显示出良好的结果。所获得的具有统计学稳健性的结构-降解速率模型可用于未来环境中有机化合物存在情况的研究，特别是它们作为水/废水处理一部分被羟基自由基降解的研究。
