Department of Biotechnology, Microbiology and Human Nutrition, Faculty of Food Science and Biotechnology, University of Life Sciences in Lublin, 8 Skromna Street, 20-704 Lublin, Poland.
Department of Analytical Chemistry, Faculty of Chemistry, Gdańsk University of Technology, 11/12G Narutowicza Street, 80-233 Gdańsk, Poland.
Molecules. 2024 Oct 1;29(19):4667. doi: 10.3390/molecules29194667.
In food authentication, it is important to compare different analytical procedures and select the best method. The aim of this study was to determine the fingerprints of Zweigelt and Rondo wines through headspace analysis using ultra-fast gas chromatography (ultra-fast GC) and to compare the effectiveness of this approach at classifying wines based on grape variety and type of malolactic fermentation (MLF) as well as its greenness and practicality with three other chromatographic methods such as headspace solid-phase microextraction/gas chromatography-mass spectrometry with carboxen-polydimethylosiloxane fiber (SPME/GC-MS with CAR/PDMS fiber), headspace solid-phase microextraction/gas chromatography-mass spectrometry with polyacrylate fiber (SPME/GC-MS with PA fiber), and ultra performance liquid chromatography-photodiode array detector-tandem mass spectrometry (UPLC-PDA-MS/MS). Principal Component Analysis (PCA) revealed that fingerprints obtained using all four chromatographic methods were suitable for classification using machine learning (ML). Random Forest (RF) and Support Vector Machines (SVM) yielded accuracies of at least 99% in the varietal classification of Zweigelt and Rondo wines and therefore proved suitable for robust fingerprinting-based Quality Assurance/Quality Control (QA/QC) procedures. In the case of wine classification by the type of MLF, the classifiers performed slightly worse, with the poorest accuracy of 91% for SVM and SPME/GC-MS with CAR/PDMS fiber, and no less than 93% for the other methods. Ultra-fast GC is the greenest and UPLC-PDA-MS/MS is the most practical of the four chromatographic methods.
在食品鉴定中，比较不同的分析程序并选择最佳方法非常重要。本研究的目的是通过顶空分析使用超快速气相色谱（ultra-fast GC）来确定 Zweigelt 和 Rondo 葡萄酒的指纹，并比较该方法在基于葡萄品种和类型的葡萄酒分类方面的有效性根据苹果酸-乳酸发酵（MLF）以及其与其他三种色谱方法（顶空固相微萃取/气相色谱-质谱联用与 CAR/PDMS 纤维（SPME/GC-MS 与 CAR/PDMS 纤维）、顶空固相微萃取/气相色谱-质谱联用与聚丙烯酸纤维（SPME/GC-MS 与 PA 纤维）和超高效液相色谱-光电二极管阵列检测器-串联质谱（UPLC-PDA-MS/MS）的绿色性和实用性。主成分分析（PCA）表明，使用所有四种色谱方法获得的指纹图谱均适合使用机器学习（ML）进行分类。随机森林（RF）和支持向量机（SVM）在 Zweigelt 和 Rondo 葡萄酒的品种分类中均获得了至少 99%的准确率，因此非常适合基于稳健指纹的质量保证/质量控制（QA/QC）程序。在 MLF 类型的葡萄酒分类方面，分类器的性能稍差，SVM 和 SPME/GC-MS 与 CAR/PDMS 纤维的准确率最差为 91%，而其他方法的准确率不低于 93%。超快速 GC 是四种色谱方法中最环保的，而 UPLC-PDA-MS/MS 是最实用的。
