Abstract

The progress in computational techniques throughout the past decade enables  the use of in silico methods to predict ADMET properties in parallel or  prior to experimental investigations. Despite the success of QSAR  approaches, it is critical to assess interactions between drug candidates and relevant proteins at the atomic level. As the three-dimensional structures of several major ADMET proteins become available like metabolizing enzymes, nuclear receptors etc., protein structure-based computations can be performed to complement or to go beyond QSAR studies. Applying docking-scoring methods to ADMET proteins  

is a challenging process because they usually have a large and flexible binding cavities. We applied protein structure-based methods to gain mechanistic insights from probing drug-like molecules binding to metabolizing enzymes and to develop automated protocols to predict potent binders for two metabolizing enzymes: the phase I, cytochrome

P450 2D6 (CYP2D6), and the phase II, sulfotransferase. We used  molecular dynamics to select suitable diverse protein conformations and  subsequent protein-ligand docking to discriminate experimentally known  binders among a large chemical compound collection for CYP2D6 and  sulfotransferase. Our results suggest that structure-based ADMET approach is useful for prioritizing compounds and may be used to go 'beyond QSAR profiling' in order to assist decision-making in drug  

discovery.