Quantitative Structure-Activity Relationship Modelling of Nigerian Plant-Derived Flavonoids as Antidiabetic Agents

Diabetes mellitus is a rapidly growing public health challenge in Nigeria, and the identification of plant-derived flavonoids with antidiabetic potential through quantitative structure-activity relationship modelling represents a powerful in silico approach that reduces the cost and time of drug candidate identification. This study performed quantitative structure-activity relationship analysis for a dataset of 45 flavonoid compounds isolated from plants used in Nigerian traditional medicine for diabetes management. Inhibitory activity against alpha-glucosidase was used as the primary biological endpoint, expressed as IC50 values. Molecular structures were optimised using density functional theory at the B3LYP/6-311G++ level. Molecular descriptors were calculated using PaDEL-Descriptor software, including topological, geometric, electronic, and constitutional descriptors. Descriptor selection was performed using genetic algorithm and forward stepwise regression. The best quantitative structure-activity relationship model was built using multiple linear regression and validated by internal cross-validation (leave-one-out) and external validation set predictions. The validated model incorporated five descriptors with statistically significant contributions, dominated by the topological polar surface area, number of hydrogen bond donors, and lipophilicity parameters. Predicted IC50 values correlated well with experimental values, with R2 of 0.912 and Q2loo of 0.878. The model revealed that low molecular flexibility, moderate lipophilicity, and specific hydrogen-bonding capacity are essential structural features for alpha-glucosidase inhibition. These findings provide actionable structural guidelines for the rational design of improved antidiabetic flavonoid candidates from Nigerian medicinal plant resources and represent an original contribution to computational medicinal chemistry in Nigeria.

Keywords: quantitative structure-activity relationship, flavonoids, antidiabetic, alpha-glucosidase, computational chemistry