Computational discovery and validation of food-derived small peptides with xanthine oxidase inhibitory activity using multidimensional descriptors

• Computational chemistry combining machine learning to predict inhibitory peptides. • Screened 168,000 peptides for XOD inhibition using docking and OneClass-SVM. • Identified AGWK and VSW as novel food-derived XOD inhibitory peptides. Hyperuricemia, largely influenced by dietary patterns and high consumption of purine-rich foods, has become an important nutritional concern associated with gout and metabolic imbalance. Xanthine oxidase (XOD), the rate-limiting enzyme in uric acid production, is a well-recognized target for lowering uric acid levels. In this context, food-derived bioactive peptides are attracting increasing interest as natural agents for uric acid regulation and functional food development. Most reported XOD-inhibitory peptides are short fragments, particularly tripeptides and tetrapeptides, consistent with the fact that dietary proteins are predominantly digested into small peptides before absorption. Building on this knowledge, we employed an integrated computational–experimental strategy to identify novel XOD-inhibitory peptides. Multidimensional descriptors were constructed as an integrated feature set of 1D/2D/3D molecular descriptors, enabling comprehensive characterization of peptide–XOD binding profiles for downstream modeling. Molecular docking revealed distinct interaction patterns across peptide lengths, highlighting GLY737, GLU664, and ARG839 as key binding residues. Based on docking outputs, multidimensional molecular descriptors were generated and used in a one-class support vector machine (OC-SVM) model to identify high-confidence candidates. Two peptides, AGWK and VSW, were experimentally validated, showing effective XOD inhibition (IC 50 = 1.24 and 1.59 mM, respectively). In vitro , both peptides lowered intracellular uric acid and modulated urate transporters. In vivo , they improved biochemical and histological markers in hyperuricemic mice. These findings demonstrate a novel approach to bioactive peptide discovery and support their potential in functional food development.