There has been a growing interest in extracting antioxidant
peptides
from food proteins. This study aimed to develop efficient computer-aided
approaches to accelerate the screening efficiency of antioxidative
dipeptides. A newly developed quantitative structure–activity
relationship model and an improved hydrolysis simulation tool, R-PeptideCutter,
were applied to screen high-activity dipeptides in sorghum kafirin.
The R
2
Test and MSETest values were 0.6082, 0.6764 and 0.5302, 0.5467, respectively, for
2,2′-azinobis(3-ethylbenzothiazoline-6-sulfonate) (ABTS) radical
scavenging capacity and oxygen radical absorbance capacity (ORAC)
models. N-terminus residues dominated the antioxidant activity, especially
in the ABTS assay, and Y and W at the N-terminus strongly corresponded
to higher activity in both assays. The dipeptide YR was predicted
as the strongest antioxidant in kafirin (3.352/2.099 μmol Trolox/μmol
peptide for ABTS/ORAC activity). Eight kafirin-derived dipeptides
were synthesized for model validation. The corresponding ORAC model
achieved greater prediction performance, while the ABTS radical scavenging
capacity model showed an underestimation in prediction. The improved
tool and knowledge can be applied to other proteins and benefit the
research and development on antioxidant peptides.