Abstract:

Generalized correlative index (GCI) is a novel user-oriented descriptor for molecular structural representation.By defining generalized correlative function (GCF), property correlative parameters (PCPs) and distance-relational function (DRF), GCI is based on molecular topological pattern and mode of apex connection and is able to perform self-adaptation and regulation in actual problems, thus possessing more extensive applications and superior correlation with properties.Applying this index to normal boiling point (NBP) of three classes of organic compounds, the resultant QSPR models all have their correlative coefficients R_cum above 0.99.It shows that GCI is superior in selecting molecular structures and correlating with molecular properties.

摘要： 将“面向用户”实际应用的观点及对目标问题“自适应性”的思想引入到分子结构表征当中。通过定义广义相关函数、性质相关参数以及距离关系函数等概念，基于分子拓扑图形特征和顶点连接方式提出了一种新型分子结构性质表征方法：广义相关指数(generalized correlative index, GCI)。尝试使用GCI指数对3类有机化合物常压沸点(NBP)进行了系统的定量构性关系(QSPR)研究，所得回归模型复相关系数皆在0.99以上。研究结果表明，GCI指数具有良好的化学结构选择性和优良的物理化学性质相关性。