Computational investigation of protein-ligand interactions in anti-diabetic agents

Abstract

Effort to apply computational power to the combined chemical and biological space in order to streamline computer aided drug discovery and development is the spirit of scientific method. India is a world capital of diabetes. Immediate attention is required for the development of a novel drug for diabetes. Though Type 2 Diabetes has many drugs, it lacks 100% effective cure. The drug design is mainly based on target identification, protein-ligand interactions and the active site residues. Our study concentrates mainly on the highly active and conserved amino acid residues of 5 important proteins responsible for causing Diabetes. An investigation has been carried out to study the mode of binding as well as the affinities of drug-like compounds from ZINC database(a free database for virtual screening), some plant compounds and chalcones as anti-diabetic agents by performing protein ligand interactions using various docking softwares. 3-dimensional structural coordinates of 1AH3 (Aldose Reductase) from protein data bank was selected for analysis based on the Root Mean Square Deviation (RMSD) and reports in literature. All default parameters are used in docking runs. A total of 1001 and 837 (search result based on range and average physico-chemical properties of co-crystallized aldose reductase ligands) compounds from ZINC database; 85 cuminum cyminum and 267 compounds from 7 different plants, 722 chalcone compounds from in-house chalcone database were docked with 1AH3 protein and the dock scores were recorded. There were many compounds which had a dock score greater than the average dock score (-126.048 kcal/mol) of the inhibitors taken from literature. ZINC00447821 (-150.707 kcal/mol) and ZINC06075556 (-186.887 kcal/mol) resulted as best compounds from 1001 and 837 hits of ZINC database.Further consensus scoring was implemented to evaluate the efficiency of docking program as a combination of these scoring functions have been shown to outperform one single scoring function.