An integrated in silico and multi-omics workflow identifies novel microbiota-derived umami peptides and elucidates T1R1/T1R3 binding mechanisms.

Microbiota-derived umami peptides show strong potential for industrial applications, yet efficient high-throughput screening remains challenging. To address this limitation, this study developed an integrated workflow combining multi-omics analysis with in silico screening. Integration of two machine-learning predictors improved screening accuracy to 91.8%. Processing-related filters, including stability, solubility, allergenicity, and toxicity, were applied to prioritize peptides with favorable physicochemical and safety profiles. Using this workflow, three novel umami peptides were identified from the umami-producing bacterium Tetragenococcus halophilus through combined transcriptomic and peptidomic analyses. Their taste thresholds (0.045-0.198 mmol/L) were validated by sensory evaluation and electronic tongue analysis. Molecular docking using an AlphaFold2-modeled T1R1/T1R3 receptor revealed key binding interactions, including hydrogen bonds involving TYR192 and ARG249. Molecular dynamics simulations further demonstrated that AHQTEGAY exhibited the most stable and compact binding conformation with T1R3, explaining its superior umami intensity and synergistic effects. This workflow provides an efficient strategy for discovering microbiota-derived umami peptides.