An efficient methodology for modeling to predict wine aroma expression based on quantitative data of volatile compounds: A case study of oak barrel-aged red wines

An efficient artificial intelligence algorithm for predicting the sensory quality of green and black teas based on the key chemical indices

The key components dominating the quality of green tea and black tea are still unclear. Here, we respectively produced green and black teas in March and June, and investigated the correlations between sensory quality and chemical compositions of dry teas by multivariate statistics, bioinformatics and artificial intelligence algorithm. The key chemical indices were screened out to establish tea sensory quality-prediction models based on the result of OPLS-DA and random forest, namely 4 flavonol glycosides of green tea and 8 indices of black tea (4 pigments, epigallocatechin, kaempferol-3-O-rhamnosyl-glucoside, ratios of caffeine/total catechins and epi/non-epi catechins). Compared with OPLS-DA and random forest, the support vector machine model had good sensory quality-prediction performance for both green tea and black tea (F1-score > 0.92), even based on the indices of fresh tea leaves. Our study explores the potential of artificial intelligence algorithm in classification and prediction of tea products with different sensory quality.

Differences in Aroma Profile of Cabernet Sauvignon Grapes and Wines from Four Plots in Jieshi Mountain Region of Eastern China

The Bohai Bay region is a famous wine-growing area in China, where the rainfall is concentrated in the summer due to the influence of the temperate semi-humid monsoon climate. As such, the vineyard terrain has a significant impact on the flavor quality of the grapes and the resulting wines. To explore the relationship between the 'Cabernet Sauvignon' wine style and terrain, this study takes four different plots in the Jieshi Mountain region to investigate the differences in the aroma profile of Cabernet Sauvignon grapes and wines of two consecutive vintages. Based on two-way ANOVA, there were 25 free and 8 glycosylated aroma compounds in the grapes and 21 and 10 aroma compounds with an odor activity value greater than 0.1 in the wines at the end of alcohol fermentation (AF) and malolactic fermentation (MLF), respectively, that varied among the four plots. Wines from the four plots showed a significant difference in floral and fruity aroma attributes, which were mainly related to esters with high odor activity values. The difference in concentration of these compounds between plots was more pronounced in 2021 than in 2020, and a similar result was shown on the Shannon-Wiener index, which represents wine aroma diversity. It has been suggested that high rainfall makes the plot effect more pronounced. Pearson's correlation analysis indicated that concentrations of (E)-3-hexen-1-ol in grapes and ethyl 3-methylbutanoate, ethyl hexanoate, isoamyl acetate, isopentanoic acid, and phenethyl acetate in wines were strongly positively correlated with the concentrations of N, P, K, Fe, and electrical conductivity in soil but negatively correlated with soil pH. This study laid a theoretical foundation for further improving the level of vineyard management and grape and wine quality in the Jieshi Mountain region.

Comparison of Chemical and Sensory Profiles between Cabernet Sauvignon and Marselan Dry Red Wines in China

The differences in chemical and sensory characteristics between Marselan and Cabernet Sauvignon in China were investigated with gas chromatography-mass spectrometry (GC-MS) and high-performance liquid chromatography-triple quadrupole mass spectrometry (HPLC-QqQ-MS/MS), combined with color parameters and sensory data. The paired t-test results showed that terpenoids, higher alcohols, and aliphatic lactones were significantly different according to the grape variety. Meanwhile, terpenoids could be considered as marker aroma compounds to distinguish Marselan wines from Cabernet Sauvignon, which could explain the distinct floral note in Marselan wines. The mean concentrations of the mv-vsol, mv-vgol, mv-vcol, mvC-vgol, mv-v(e)cat, mvC-v(e)cat, mv-di(e)cat, and cafA were higher in Marselan wines than Cabernet Sauvignon wines, and these compounds might confer Marselan wines with a deeper color, more red hue, and higher tannin quality. The phenolic profiles of Marselan and Cabernet Sauvignon wines were influenced by the winemaking process, mitigating the varietal differences. As for sensory evaluation, the intensities of herbaceous, oak, and astringency of Cabernet Sauvignon were more pronounced than Marselan, whereas the Marselan wines were characterized by a high color intensity and more redness, together with floral, sweet, and roasted sweet potato attributes, and tannin roughness.