Data-Driven Prediction of Molecular Biotransformations in Food Fermentation

Fermentationproducts, together with food components, determinethe sense, nutrition, and safety of fermented foods. Traditional methodsof fermentation product identification are time-consuming and cumbersome,which cannot meet the increasing need for the identification of theextensive bioactive metabolites produced during food fermentation.Hence, we propose a data-driven integrated platform (FFExplorer, http://www.rxnfinder.org/ffexplorer/) based on machine learning and data on 2,192,862 microbial sequence-encodedenzymes for computational prediction of fermentation products. UsingFFExplorer, we explained the mechanism behind the disappearance ofspicy taste during pepper fermentation and evaluated the detoxificationeffects of microbial fermentation for common food contaminants. FFExplorerwill provide a valuable reference for inferring bioactive "darkmatter" in fermented foods and exploring the application potentialof microorganisms.