摘要
基于13种原子类型的分子电性距离矢量(M_(t))表征37个苯氧基苯甲酰胺衍生物的分子结构,并与其对葡萄灰霉病菌的杀菌活性(B_(C))关联。通过最佳变量子集回归方法建立上述化合物杀菌活性的三参数(M_(15)、M_(85)、M_(78))的定量构效关系(quantitative structure-activity relationship,QSAR)模型。其相关系数(r)和标准偏差(S_(D))分别为0.891和13.28。通过r、r^(2)_(cv)、F、V_(IF)、A_(IC)、F_(IT)等检验,上述模型具有令人满意的相关性、稳健性和预测能力。结果显示—CH_(2)—、—C、—OH、—O—和—X(如—F、—Cl)等分子结构单元直接影响这些化合物的杀菌活性。将M_(15)、M_(85)、M_(78)作为人工神经网络的输入层结点,采用3∶4∶1的网络结构,利用BP算法获得了一个令人满意的B_(C)模型,其r^(2)和S_(D)分别为0.986和3.36,表明B_(C)与三参数呈现优异的非线性关系。
The molecular structure of 37 phenoxybenzamide derivatives was characterized based on the molecular electronegativity distance vector(M_(t))of 13 atom types,and was associated with their fungicidal activity(B_(C))a gainst Botrytis cinerea.A three-parameter(M_(15),M_(85),M_(78))quantitative structure-activity relationship(QSAR)model for the fungicidal activity of the above compounds was established by the best variable subset regression method.Its correlation coefficient(r)and standard deviation(S_(D))were 0.891 and 13.28,respectively.Through tests such as r,r^(2)_(cv),F,V_(IF),A_(IC),and F_(IT),the above model has satisfactory correlation,robustness,and predictive ability.The results show that molecular structural units such as-CH_(2)-,-C,-OH,-O-,and-X(e.g.,-F,-Cl)directly affect the fungicidal activity of these compounds.M_(15),M_(85),and M_(78)were used as the input layer nodes of the artificial neural network,and a 3∶4∶1 network structure was adopted.A satisfactory B_(C)model was o btained using the BP algorithm,with r^(2) and standard deviation S_(D)of 0.986 and 3.36,respectively,indicating an excellent nonlinear relationship between B_(C)and the three parameters.
作者
冯惠
李靖
石春玲
冯长君
FENG Hui;LI Jing;SHI Chunling;FENG Changjun(School of Material and Chemical Engineering,Xuzhou University of Technology,Xuzhou 221018,China)
出处
《生态毒理学报》
北大核心
2025年第6期414-420,共7页
Asian Journal of Ecotoxicology
基金
国家自然科学基金资助项目(21075138)
结构化学国家重点实验室开放基金资助项目(2016028)。
关键词
苯氧基苯甲酰胺衍生物
杀菌活性
电性距离矢量
人工神经网络
定量构效关系(QSAR)
phenoxybenzamide derivative
antifungal activity
electronegativity distance vector
artificial neural n etwork
quantitative structure-activity relationship(QSAR)
