Exploring details about structure requirements based on antioxidant tripeptide derived from β-Lactoglobulin by in silico approaches

β-Lactoglobulin is one of the proteins in milk possessing antioxidant activity. The peptides derived from β-Lactoglobulin exhibit higher antioxidant activities than the most commonly used antioxidant. Furthermore, the detailed structure–activity relationship of these antioxidant peptides has not been elucidated. Therefore, in the present work, two-dimensional quantitative structure–activity relationship (2D-QSAR) and three-dimensional quantitative structure–activity relationship (3D-QSAR) models were constructed to investigate the structural factors affecting activities and gave information for the rational design of novel antioxidant peptides. After calculation and screening of molecular descriptors, linear and nonlinear models were developed by multiple linear regression (MLR), partial least squares regression (PLSR) and support vector machines (SVM) approaches. The statistical parameters are as follows: R² = 0.643, Q² = 0.553/MLR, R² = 0.612, Q² = 0.5278/PLSR, R² = 0.7085, Q² = 0.6887/SVM, indicating that the SVM model is superior to the MLR and PLSR models. In addition, in the 3D-QSAR models, the Dragon-CoMFA (R²_cv = 0.537, R²_pred = 0.5201) and Dragon-CoMSIA (R²_cv = 0.665, R²_pred = 0.6489) methods were conducted to predict the antioxidant activities. Comparison of statistical parameters illustrates that the suitability of Dragon-CoMSIA is superior to the Dragon-CoMFA model. The results show the robustness and excellent prediction of the proposed models, and would be applied for modifying and designing novel and potent antioxidant peptides.