Department of Chemistry, Faculty of Science, University of Isfahan, Isfahan 81746-73441, Iran.
Department of Chemistry, Sharif University of Technology, P.O. Box 11155-3516, Tehran, Iran.
J Chromatogr A. 2012 Aug 17;1251:176-187. doi: 10.1016/j.chroma.2012.06.011. Epub 2012 Jun 15.
Multivariate curve resolution (MCR) and multivariate clustering methods along with other chemometric methods are proposed to improve the analysis of gas chromatography-mass spectrometry (GC-MS) fingerprints of secondary metabolites in citrus fruits peels. In this way, chromatographic problems such as baseline/background contribution, low S/N peaks, asymmetric peaks, retention time shifts, and co-elution (overlapped and embedded peaks) occurred during GC-MS analysis of chromatographic fingerprints are solved using the proposed strategy. In this study, first, informative GC-MS fingerprints of citrus secondary metabolites are generated and then, whole data sets are segmented to some chromatographic regions. Each chromatographic segment for eighteen samples is column-wise augmented with m/z values as common mode to preserve bilinear model assumption needed for MCR analysis. Extended multivariate curve resolution alternating least squares (MCR-ALS) is used to obtain pure elution and mass spectral profiles for the components present in each chromatographic segment as well as their relative concentrations. After finding the best MCR-ALS model, the relative concentrations for resolved components are examined using principal component analysis (PCA) and k-nearest neighbor (KNN) clustering methods to explore similarities and dissimilarities among different citrus samples according to their secondary metabolites. In general, four clear-cut clusters are determined and the chemical markers (chemotypes) responsible to this differentiation are characterized by subsequent discriminate analysis using counter-propagation artificial neural network (CPANN) method. It is concluded that the use of proposed strategy is a more reliable and faster way for the analysis of large data sets like chromatographic fingerprints of natural products compared to conventional methods.
多元曲线分辨（MCR）和多元聚类方法以及其他化学计量学方法被提出用于改进柑橘果皮次生代谢产物的气相色谱-质谱（GC-MS）指纹图谱的分析。通过这种方式，可以解决 GC-MS 分析指纹图谱时出现的色谱问题，如基线/背景贡献、低 S/N 峰、不对称峰、保留时间漂移以及共洗脱（重叠和嵌入峰）。在本研究中，首先生成了柑橘次生代谢物的信息丰富的 GC-MS 指纹图谱，然后将整个数据集分段到一些色谱区域。对于十八个样本的每个色谱段，以列方式增加 m/z 值作为公共模式，以保留用于 MCR 分析的双线性模型假设。扩展多元曲线分辨交替最小二乘法（MCR-ALS）用于获得每个色谱段中存在的组分的纯洗脱和质谱轮廓及其相对浓度。找到最佳的 MCR-ALS 模型后，使用主成分分析（PCA）和 k-最近邻（KNN）聚类方法检查解析成分的相对浓度，根据次生代谢物探索不同柑橘样品之间的相似性和差异性。总体上，确定了四个明显的聚类，并用反向传播人工神经网络（CPANN）方法进行判别分析来表征负责这种分化的化学标志物（化学型）。结果表明，与传统方法相比，该策略的使用是分析像天然产物的色谱指纹图谱这样的大数据集的更可靠和更快的方法。
