AI and the World's Hard Problems · Life sciences · Nature (computational biology) · 2026-10-20
AI-guided discovery of fungal-derived salty and saltiness-enhancing peptides for sodium-reduced foods
To enable sodium reduction in foods, the study established an integrated AI-guided workflow combining in silico hydrolysis, embedding-based machine learning, electronic tongue validation, cellular evaluation, molecular docking, and molecular dynamics simulation to identify fungal-derived saltiness-modulating peptides. ESM-3 embeddings showed the best classification performance and were used to screen peptides generated from 14 fungal proteomes, yielding six synthesized candidates. Electronic tongue analysis identified AHDQY as the strongest salty peptide and ANSDL as the most effective saltiness enhancer in a low-salt NaCl system, and the peptides showed good cytocompatibility and protected HUVECs against high-salt-induced cellular injury.