Prediction Of Sensory Evaluation And Metabolomic Analysis In Industrial Beer Fermentation Process

In this study, experimental data from industrial fermentation was adopted to create predictive models for beer sensory evaluation and analyze the association between the flavor compounds and sensory evaluation for the purpose of industrial applications. Meanwhile, investigation of Lager yeast metabolomics analysis was carried out, which is of great significance to optimize the flavor components and reveal the underlying laws of the formation of esters and alcohols in beer fermentation.Data from five-hundreds of samples of commercial Tsingtao beers of different production batches were provided, which consists of flavor index and sensory score. The regression model of the relationship between flavor compounds and sensory evaluation was established by partial least squares (PLS), genetic algorithm back-propagation neural network (ANN-BP) and support vector machine (SVM). It was shown that SVM with a Radial Basis Function (RBF) had a better performance of prediction accuracy for validation set (96.2%) than other models. Relatively lower prediction abilities were observed for ANN-BP (52.1%) and PLS (42.7%). In addition, the kernel function of SVM played an essential role of model training when the prediction accuracy of SVM with polynomial kernel function was 32.9%.For the industrial fermentation processes, Lager yeast intracellular metabolites were identified, a total of 62 kinds of intracellular metabolites was qualitatived and quantitatived. The intracellular metabolites included amino acids, sugar alcohols, sugars, sugar acids, organic acids and fatty acids, among which the most abundant metabolites were amino acids (17 kinds). Here, principal component analysis (PCA) and PLS for high-throughput data were used. PCA results showed that phosphoric acid, trehalose, succinic acid, glutamine, aspartic acid and alanine contributed to the principal components significantly, which means that these metabolites varied obviously in different stages of fermentation. PLS model was applied to quantify the impact of cellular metabolites on flavor substances. There are 14 cellular metabolites’VIP value> 1, which means that these cellular metabolites are key metabolites. The greatest impactive factor on flavor metabolites were mainly intracellular amino acids. ISome other key metabolites such as succinic acid (VIP=1.3465) and metabolites associated with fatty acid include decanoate (VIP=1.3025)and palmitic acid (VIP=1.2915) also haves influence on flavor substances. n addition, phosphate (VIP=1.2888) and trehalose (VIP=1.2888) also have an impact on the flavor characteristics.