Machine learning-driven GC-MS analysis for assessing the contribution of amino acids in the flavor of cheese maturing at high temperature.

Abstract

Branched-chain (BCAA) and aromatic amino acids (AAA) are precursors of volatile compounds with pleasant flavors in cheese. In this study, BCAA and AAA were added to cheese which matured at 30 °C for 7 days to assess their effects on accelerating cheese ripening. Machine learning applied to GC-MS data enabled a comprehensive and holistic assessment of ripening acceleration, effectively integrating multiple volatile compounds to overcome the limitations of traditional single-marker maturity indices. This analysis revealed that flavor components of Ile and Tyr cheeses likely correspond to 3 months maturity, as well as Leu and Val cheeses potentially reach 6 months maturity. The combined results of the number of Lactococcus, hydrolysis degree, FAA composition revealed that Leu, Tyr and Trp would promote the formation of related flavor substances through the increase of substrate content and Leu can increase the tolerance of Lactococcus in cheese. Val could be used by Lactococcus as nutritional factors to increase the numbers and tolerance of Lactococcus, further affected the production of some volatile compounds. This study proved that adding amino acids at higher ripening temperature can accelerate cheese local ripening and its possible mechanism, which provided an innovative scheme for rapid ripening cheese.