高效液相色谱-飞行时间-串联质谱和随机森林算法的蜜炙百合与生百合指纹图谱研究

Fingerprint Study of Candied Lilium and Raw Lilium Based by High Performance Liquid Chromatography-Time of Flight-Tandem Mass Spectrometry Coupling with Forest Algorithm

采用指纹图谱结合化学计量学对不同产地蜜炙百合与生百合质量进行评价。收集23批不同来源的蜜百合与生百合进行高效液相色谱(HPLC)分析,运用中药色谱指纹图谱相似度评价软件(2004药典版)建立两类百合对照指纹图谱,采用随机森林算法对其进行判别分析,筛选其主要差异性组分;并利用HPLC-飞行时间-串联质谱(TOF-MS/MS)法对差异峰进行鉴别。结果显示23批不同来源蜜炙百合与生百合共获得13个共有峰;利用随机森林算法准确地分辨了蜜炙百合与生百合,发现王百合苷D和王百合苷B对分辨两类百合具有重要贡献。

In this work,we establish the fingerprints of candied Lilium and raw Lilium from different producing areas for quality control and resource evaluation of Lilium by combining chemometrics algorithm.Twenty-three batches of candied Lilium and raw Lilium from different sources were collected for HPLC analysis.The control fingerprint of Lilium was established by using the chromatographic fingerprint similarity evaluation software(2004 Pharmacopoeia).The random forest algorithm was used to discriminate and the main differential components,which were then identified by high performance liquid chromatography-time of flight-tandem mass spectrometry(HPLC-TOF-MS/MS).The HPLC fingerprints of 23 batches of candied Lilium and raw Lilium from different sources were established,and 13 common peaks were obtained.The random forest algorithm was used to accurately distinguish the candied Lilium and the raw Lilium,and was found to have an important contribution to the discrimination of the two types of Lilium.Regaloside D and Regaloside B.The results indicate that the combination of fingerprint and pattern recognition methods can provide a favorable basis for Lilium quality control and evaluation.