摘要
采用Cerius2软件中的主成分分析法和Var.Jarvis-Patrick聚类方法对新型2-烷氧基-4(3H)-喹唑啉酮类化合物进行分类,然后用遗传算法(GFA)建立杀菌活性与优选的分子结构描述符之间的构效关系(QSAR)模型,所建模型均通过显著性检验,交叉验证系数(CV-r2)最大达0.996,具有良好的预测可靠性。结果表明:分子空间性质、给电子能力的大小和热力学性质等是影响活性的主要因素,并用所建最优模型预测了5个经典生物电子等排取代物的杀菌活性。该研究可为新型杀菌剂先导化合物的设计与合成提供理论指导。
This article presented the study of novel 2-alkoxy-4（3H）-quinazolinones derivatives from literature sorted by Var. Jarvis-Patrick clustering method coupled with principal component analysis（PCA） and built 2D-QSAR models between fungicidal activity and their selected molecular structure descriptors based on the genetic function approximation（GFA） using Cerius^2 4.10 software. The models were all significant with the maximal cross-validated r^2 of 0.996 accounting for good predictive abilities and than used to predict five compounds constructed according to classical bioisoteric replacement with unreported fungicidal activity. The results show the activities are greatly related to molecular steric, electronic and thermodynamic property. This study can provide academic guidance of structural design and synthesis for novel fungicidal reagents with high activities.
出处
《农药》
CAS
北大核心
2008年第5期339-343,共5页
Agrochemicals
基金
国家自然科学基金资助项目(30500339)
浙江省自然科学基金资助项目(Y407308)
