紫丁香花精油挥发性成分色谱保留指数的预测模型

Predictable model of chromatographic retention index of volatile in essential oil from flowers of Syringa oblata Lindl.

定量结构-色谱保留指数相关性(QSRR)方法已成为色谱领域中的有用工具,用于解释与预测各种物质的色谱行为。在分子拓扑理论的基础上,计算了64种紫杉醇花精油挥发性成分化合物的三类拓扑指数值:价分子连接性指数(χi),电性拓扑状态指数(ek)和分子电性距离矢量(mj),并随机选取其中的51个作为训练集,其他分子做检验集,采用多元回归中逐步筛选的方法建立了分子结构和色谱保留指数的最优预测模型(R=0.982),用检验集检验了所建模型的预测能力,并经Jackknife检验,交互检验和外部检验验证了模型的总体稳健性。

Quantitative structure-retention relationships（QSRR） have been demonstrated to be a powerful tool in chromato-graphy,which have been used to explain and predict the chromatographic behavior of various classes of substances.Based on the chemical topological theory,molecular connectivity index（χi）,electrotopological state index（ek） and electro-negativity distance vector（mj） of 64 volatile compounds in essential oil from flowers of Syringa oblata Lindl were calculated.Selecting randomLy 51 compounds as training set and the rest as test set,the best QSRR predictable model was built by stepwise of multivariate liner regression analysis method（R=0.982）,whose predictability was validated by the test set.And this model exhibited excellent stability by Jackknife method,LOO cross-validation and external-validation procedure.