Related Experiment Video
Updated: Jan 12, 2026
Laser Microdissection-Based Protocol for the LC-MS/MS Analysis of the Proteomic Profile of Neuromelanin Granules
Published on: December 16, 2021
Machine learning-driven screening of antioxidant peptides from macadamia nuts: In vitro experimental validation and
Weiye Jiang1, Zehui Zhu2, Fei Pan3
1Key Laboratory of Geriatric Nutrition and Health, Beijing Technology and Business University, Ministry of Education, Beijing 100048, China; Beijing Engineering and Technology Research Center of Food Additives, Beijing Technology and Business University, Beijing 100048, China.
Abstract:
Oxidative stress contributes to the pathogenesis of chronic diseases, driving the need for effective and natural antioxidant agents. Antioxidant peptides are promising alternatives; however, their conventional identification methods are labor-intensive and inefficient. This study presents a machine learning (ML)-based framework to accelerate the discovery of antioxidant peptides from macadamia nut proteins. Sequence embeddings from ESM-2 were combined with ten ML algorithms to construct binary classification models for four antioxidant assays (ABTS, DPPH, ORAC, FRAP), each achieving predictive accuracies above 92 %. Top-performing models, including XGBoost for ABTS and SVC for DPPH, were applied to screen peptides from macadamia nut proteins, leading to the identification of SYLDL. In vitro analysis showed that antioxidant activities of DPPH, FRAP, and ORAC assays peaked after 8 h of hydrolysis, while ABTS activity reached its maximum (68.7 %) at 10 h. The peptide SYLDL was synthesized and evaluated in a hydrogen peroxide-induced oxidative stress model using HepaRG cells. At 75 μg/mL, SYLDL treatment resulted in a statistically significant increase in cell viability and total antioxidant capacity, along with reductions in intracellular ROS and MDA) levels (p < 0.05. SYLDL also significantly elevated levels of GSH and CAT activity compared to the control group. Western blot analysis confirmed the upregulation of antioxidant-related proteins HO-1(Heme Oxygenase-1), Keap1(Kelch-like ECH-associated protein 1), NQO1(NAD(P)H: quinone oxidoreductase 1), and Nrf2(Nuclear factor-E2-related factor 2) following SYLDL treatment (p < 0.05). Furthermore, frontier molecular orbital calculations indicated a low energy gap, consistent with strong electron-donating potential and antioxidant activity. Overall, these findings support the potential of SYLDL as a candidate antioxidant peptide, and demonstrate the utility of integrating ML-based screening with experimental validation in the discovery of bioactive food-derived peptides.
