Multidimensional strategy for discovering saltiness-enhancing peptides in shrimp heads integrating ultra-high pressure hydrolysis and machine learning.

This study aims to develop a comprehensive strategy to investigate whether the integration of ultra-high pressure (UHP)-assisted enzymatic hydrolysis with machine learning and molecular docking can effectively identify salty peptides (SPs) from Litopenaeus vannamei heads. UHP treatment (400 MPa, 30 min) significantly enhanced proteolysis, yielding more low-molecular-weight peptides (<3 kDa, 66.44 %) than conventional methods. Peptidomics identified 211 differentially expressed peptides enriched in hydrophobic/acidic residues (e.g., Asp, Glu) linked to saltiness. A machine learning model integrating amphipathic pseudo-amino acid composition (APAAC) and ensemble classifiers (KNN-LR-RF) achieved superior predictive performance for SP identification. Molecular docking revealed five novel peptides (DDL, TVT, DPS, VM, PM) with strong binding affinities (-5.103 to -6.251 kcal/mol) to the TMC4 receptor, validated via solid-phase synthesis and electronic tongue analysis. This work provides a sustainable approach for valorizing shrimp processing waste and advancing low-sodium food ingredient development through bioactive peptide discovery.