Difference of volatile flavor components of typical distilled spirits from China and Hungaria based on flavor omics and machine learning model

In this study, the volatile flavor components of different types of Baijiu and Palinka (from Hungaria) with different flavors were analyzed using headspace solid-phase microextraction-gas chromatography-mass spectrometry (HS-SPME-GC-MS). Combining multivariate statistical methods and machine learning models, the characteristic flavor components of two distilled spirits were systematically analyzed. The model was constructed using orthogonal partial least squares discriminant analysis (OPLS-DA), and differential flavor substances were screened using variance analysis and variable important in projection (VIP) values. The results showed that a total of 80 volatile flavor substances were identified in the 2 types of distilled spirits, including 2 alcohols, 7 ethers, 3 acids, 11 terpenes, 10 ketones, 17 esters, and 30 others. Combining variance analysis and VIP values, a total of 25 differential flavor substances (VIP>1, P<0.05) were screened out. Among them, terpenes and ethers were the main differential characteristic components of Palinka, while esters and alcohols were the main differential substances of Baijiu. The classification model based on the random forest algorithm indicated that triethyl orthoformate and (+)-limonene and other substances were the key components for distinguishing the 2 types of distilled spirits. The research results provided a scientific basis for revealing the influence of different fermentation raw materials and processes on the flavor formation of distilled spirits.