本文创新性地将SPME-GC-MS检测手段和代谢组学数据处理技术结合应用于白酒特征化合物鉴定和真假区分中。17个白酒样品经顶空固相微萃取，富集酒中挥发性化合物。所得GC-MS数据集经代谢组学技术降维处理，经聚类分析真酒和假酒样品得到了正确的区分，同时其他酒系列所得分类结果基本符合实际酒样信息。在主成分分析中，提取了PC1和PC2两个主成分，解释了不同酒样品特征变量58.7%的方差信息，其中PC1为44.7%，真假酒霍特林椭圆区域区分明显，可视化效果直观。其他系列酒样也保持了对真实酒样信息的吻合。利用偏最小二乘判别分析法建立酒类相关模型，去除基酒样品以突出真假酒的物质区别，得出变量重要性表并且查库得出相关物质十二种，差异最显著的前三位特征化合物是乙酸丁酯、己酸异戊酯和己酸-1-甲乙酯。

The headspace solid-phase microextraction-gas chromatography-mass spectrometry (HS-SPME-GC-MS) detection method and metabolomics data processing technology were creatively applied in the identification of characteristic compounds in liquor and the assessment of liquor adulteration. Seventeen liquor samples were processed through headspace solid phase microextraction, and the volatile compounds in the liquor samples were accumulated. The dimension reduction of the GC-MS data set was achieved using the metabolomics technique. The clustering analysis was employed to correctly distinguish between contaminated and authentic liquors. The classification results from other series of liquors were generally consistent with actual liquor samples. In a principal component analysis, PC1 and PC2 were extracted, and accounted for 58.7% of the variance in the characteristic variables of different liquor samples. Among them, 44.7% of variance was explained by PC1, and a clear distinction was observed between the Hotelling’s elliptical regions of authentic and contaminated liquor samples. The information for authentic liquor samples was maintained in other samples. A liquor related model was established based on a partial least squares discriminant analysis for the removal of base liquor samples, so that the substance difference between authentic and contaminated liquor samples became significant. A variable importance table was generated and 12 related substances were obtained from the database. The top three characteristic compounds in terms of significant difference were butyl acetate, isoamyl caproate, and hexanoic acid isopropyl ester.

国家自然科学基金资助项目（81341082）；上海市科委工程中心建设（11DZ2280300）