波谱学杂志 ›› 2008, Vol. 25 ›› Issue (2): 228-233.

• 研究论文 • 上一篇    下一篇

吖啶酮生物碱的核磁共振碳谱模拟研究

何留; 梅虎; 李志良   

  1. 重庆大学 化学化工学院,重庆 400044
  • 收稿日期:2007-05-21 修回日期:2007-06-26 出版日期:2008-06-05 发布日期:2009-12-05
  • 通讯作者: 李志良

Estimation and Prediction of 13C Chemical Shifts of Acridone Alkaloids: A Quantitative Structure-Spectrum Relationship

HE Liu; MEI Hu; LI Zhi-liang   

  1. College of Chemistry and Chemical Engineering, Chongqing University, Chongqing 400044, China
  • Received:2007-05-21 Revised:2007-06-26 Online:2008-06-05 Published:2009-12-05
  • Contact: Li Zhi-liang

摘要: 采用表征分子内部化学微环境及原子所处杂化状态的结构描述子:原子电性作用矢量(AEIV)和原子杂化状态指数(AHSI),对42个吖啶酮生物碱分子792个共振碳原子进行结构表征,以多元线性回归技术建立13C核磁共振化学位移定量结构波谱关系模型,所得回归模型的复相关系数为R=0.957,标准偏差为SD=12.247. 采用留一法交互检验的结果为Rcv=0.956,标准偏差SDcv=12.331. 对外部样本集预测结果表明,AEIV和AHSI具有表征能力强、物化意义明确等优点,所建模型具有良好的稳定性和估计能力.

关键词: 13C核磁共振波谱模拟, 原子电性作用矢量, 原子杂化状态指数, 吖啶酮

Abstract: A multiple linear regression model for calculating 13C chemical shifts in acridone alkaloids from the structural indices namely the atomic electronegative interaction (AEIV) vector and the atomic hybridation state index (AHSI) was established. 13C chemical shifts of 792 equivalent carbon atoms in 42 acridone alkaloids were calculated based on their AEIV and AHSI indices. The correlation coefficients and standard deviation of modeling estimation and leave-one-out cross-validation are 0.957, 0.956 and 12.247,12.331, respectively. An external prediction set was applied to validate the prediction capabilities of the model. The results show that both AEIV and AHSI are excellent topological indices with satisfactory estimation stability and favorable predictive ability.

Key words: 13C nuclear magnetic resonance, atomic electronegativity interaction vector, atomic hybridization state index, acridone alkaloid