Modeling of Hydrophobic Interactions in Computational Ligand Screening: A Template Matching Approach

Computational ligand screening is an approach to identify potential protein ligands efficiently using computational rather than experimental techniques. One approach is to test for the ability to identify and dock true ligands from a large number of molecular structures. Using this approach as implemented in our software SLIDE(1), screenings of 100,000-plus molecules can be achieved in a few hours to a day, with each molecule taking less than one second to screen and dock. SLIDE does this screening while including both ligand and protein side-chain flexibility and water-mediated interactions, with a template-matching approach based points of complementary chemistry between the protein and each ligand candidate. The description of the protein template, representing the binding site, involves points where favorable hydrophobic or hydrogen bonding interactions can be made. Here, we present an improved method of modeling the hydrophobic interaction component of the SLIDE screening method. Protein template points are identified by representing the major patches of hydrophobic surface on the protein with a uniform density of points, while not increasing the number of points in the protein binding site. Hydrophobic interaction points on the ligand are assigned using rules which assign points at a similar density and spacing to protein template points, improving the quality of ligands dockings, especially for predominantly hydrophobic ligands. We present use of this new method on several test cases, showing docking of known ligands closer to the orientation found in the X-ray crystallographic structure and faster screening times on a test database. Combined with the new hydrogen-bond template design developed by Maria Zavodszky (see related poster), the new knowledge-based hydrophobic and hydrogen-bond template method allows unbiased screening of ligands while considering the shape and chemistry of the protein target. Abstract