In Silico Design and Atom-Based 3D-QSAR Studies of Novel Flavonoid Derivatives as Potent alpha-Glucosidase Inhibitors

This study aimed to identify the key structural features that optimize the inhibitory activity of a series of 28 flavonoid derivatives against α-glucosidase, with the ultimate goal of providing rational guidelines for the generation of highly potent drug candidates. An integrated in silico computational workflow was successfully applied. 2D chemical structures were sketched using ChemDraw Ultra and prepared via the LigPrep module using the OPLS-2005 force field to determine proper ionization and tautomeric states. An atom-based 3D-QSAR model was subsequently developed within the Phase module using partial least squares (PLS) regression to analyze 3D contour maps (comprising hydrogen-bond donor, hydrophobic, and electrostatic effects). To validate the structural insights, molecular docking simulations were conducted using the Glide extra-precision (XP) module to analyze receptor-ligand interaction profiles and binding affinities within the catalytic pocket of -glucosidase. Finally, pharmacokinetic and drug-likeness profiles were predicted using the QikProp module. The developed atom-based 3D-QSAR model demonstrated strong statistical robustness and high predictive reliability, yielding an excellent correlation coefficient for the training set (R²= 0.97), a solid cross-validated coefficient (Q²_CV= 0.69), and a high correlation coefficient for the test set (R²= 0.88) using three PLS-components. Visualization of the phase contour maps for the active compounds revealed that structural modifications on the chromone and phenyl rings, particularly the introduction of long-chain alkoxy substituents or nitrate moieties, as observed in the dataset, significantly modulate binding affinity. Glide XP docking corroborated these results, revealing favorable binding scores and consistent stabilizing interactions, such as hydrogen bonds and -stacking, with key active-site residues.