Artificial Intelligence-Aided In Silico Screening of Syzygium polyanthum Phytochemicals for Antidiabetic Drug Discovery Using ACO (Ant Colony Optimization) Algorithm

Abstrak

This research employs an artificial intelligence (AI)-driven molecular docking approach to identify potential antidiabetic compounds from Syzygium polyanthum phytochemicals targeting the α-glucosidase enzyme. The docking simulations were conducted using the PLANTS software, which utilizes an ant colony optimization (ACO) algorithm, a nature-inspired AI technique that mimics the foraging behavior of ants to explore ligand binding conformations efficiently. PLANTS integrates multiple empirical scoring functions, including ChemPLP, to evaluate protein-ligand interactions by modeling steric complementarity, hydrogen bonding, and torsional potentials, enabling accurate prediction of binding affinities. The protein structure with PDB code 2JKE was validated with a root-mean-square deviation (RMSD) of 0.2912 Å, confirming the reliability of the docking protocol. Screening results revealed seven phytochemical compounds Hexadecanoic acid 2-hydroxy-1-(hydroxymethyl), Methyl oleate, Methyl palmitate, Phytol, 9,12,15-Octadecatrien-1-ol, Nerolidol, and Eicosane exhibited lower docking scores (-96.2919 to -80.5188) than both the reference drug miglitol (-80.2642) and the native ligand (-77.2910), indicating stronger and more stable binding to the α-glucosidase active site. These findings suggest that the identified compounds have superior theoretical inhibitory potential compared to miglitol, a clinically used α-glucosidase inhibitor. The AI-based in silico screening using PLANTS thus provides a powerful, cost-effective strategy for accelerating antidiabetic drug discovery by prioritizing promising natural compounds for further experimental validation.