Bioscope Group, Physical Chemistry Department, Science Faculty, University of Vigo, Ourense, Spain.
Talanta. 2012 Mar 15;91:72-6. doi: 10.1016/j.talanta.2012.01.017. Epub 2012 Jan 14.
The variables affecting the direct matrix assisted laser desorption ionization mass spectrometry-based analysis of wine for classification purposes have been studied. The type of matrix, the number of bottles of wine, the number of technical replicates and the number of spots used for the sample analysis have been carefully assessed to obtain the best classification possible. Ten different algorithms have been assessed as classification tools using the experimental data collected after the analysis of fourteen types of wine. The best matrix was found to be α-Cyano with a sample to matrix ratio of 1:0.75. To correctly classify the wines, profiling a minimum of five bottles per type of wine is suggested, with a minimum of three MALDI spot replicates for each bottle. The best algorithm to classify the wines was found to be Bayes Net.
已研究了影响葡萄酒直接基质辅助激光解吸电离质谱分析（MALDI-MS）以用于分类目的的各种变量。仔细评估了基质的类型、葡萄酒瓶数、技术重复次数和用于样品分析的斑点数量，以获得最佳的分类效果。使用分析 14 种不同类型葡萄酒后收集的实验数据，评估了 10 种不同的算法作为分类工具。结果发现，最佳基质是α-氰基，样品与基质的比例为 1:0.75。为了正确分类葡萄酒，建议每种类型的葡萄酒分析至少 5 瓶，每瓶至少有 3 个 MALDI 斑点重复。结果发现，用于分类葡萄酒的最佳算法是贝叶斯网络。
