有机化学品与臭氧反应速率常数的定量预测模型研究

    Study on Quantitative Structure-Activity Relationship Model for Ozone Radical Reaction Constant of Organic Chemicals

    • 摘要: 有机化学品与臭氧反应速率常数(KO3)是表征有机污染物环境归趋,评估其大气环境持久性的重要参数。采用量子化学方法对不同类别的152种化合物进行结构优化,基于8个Dragon分子结构描述符,运用遗传算法构建化学品与臭氧反应速率常数的定量结构-活性关系(QSAR)模型,并根据经济合作与发展组织关于QSAR模型构建与验证导则对模型进行表征。结果显示,构建的QSAR模型经自由度校正的决定系数(Radj2)、均方根误差(RMSE)和留一法交叉验证系数(QLOO2)分别为0.784、1.127和0.744,说明模型拟合优度和稳健性均良好。外部检验参数(QEXT2)、验证集相关系数(REXT2)、验证集均方根误差(ERMS,EXT)等外部验证决定系数分别为0.664、0.761和1.039,说明模型预测能力较好。基于Williams图定义的模型应用域(AD)结果表明,模型只有1个X离群点。因此,所构建的QSAR模型可用于预测应用域内其他化学品的臭氧反应速率。

       

      Abstract: The reaction rate constants of ozone radical (KO3) is of vital importance to the characterization of the environmental fate and the evaluation of the persistence of organic pollutants in the atmosphere. In this study, quantum chemistry method was used for the structure optimization of 152 organic chemicals. Based on eight Dragon descriptors, the QSAR model was developed by employing genetic algorithm-multiple linear regressions (GA-MLR) and the model quality was evaluated according to the QSAR model development and validation guideline issued by Organization for Economic Cooperation and Development (OECD). The results show that the adjust determination coefficient (Radj2), the root mean square error (RMSE), leave-one-out cross-validation squared correlation coefficient (QLOO2) were 0.784, 1.127, and 0.744, respectively, indicating that the model has high goodness-of-fit and robustness. The external validation coefficient QEXT2, REXT2 and ERMS, EXT were 0.664, 0.761 and 1.039, respectively, indicating that the model has good predictive ability. The Williams plot were used to visualize the applicability domain (AD) of the model and the result indicate that there was only one X outlier. Thus, the QSAR model can be employed to predict -lg KO3 values of other compounds within the AD of the model.

       

    /

    返回文章
    返回