A rapid selection strategy for umami peptide screening based on machine learning and molecular docking.

ABSTRACT

Umami peptides have been the focus of umami studies in recent years because of their high nutritional value and flavor activity. However, the existing screening methods of umami peptides were cumbersome, complex, time-consuming and laborious, and it was difficult to achieve high-throughput screening. In this study, a novel umami peptide rapid screening model was designed and by using lamb bone aqueous extract as raw material, through the step-by-step screening of peptidomics, machine learning methods, and molecular docking technology. Results showed that six novel peptides about lamb bones were obtained, which verified the feasibility of the model and could be used for high-throughput screening of umami peptides. Results of molecular docking between umami peptide and T1R3 subunit revealed that the main interaction forces were hydrogen bonding and electrostatic interaction, and the key binding sites were GLU277 and SER146. It provides the basis for studying the binding mechanism of umami peptide.