Design and Experimental Testing of New Docking Methods

University Of California, San Francisco, San Francisco CA

Investigators

Linked publications & trials

Abstract

DESCRIPTION (provided by applicant): Structure-based screening using molecular docking is now widely used for ligand discovery. The technique makes many approximations and, although, it has had noteworthy successes, it is neither fully reliable nor do we understand why it fails when it does. Here, we develop model experimental systems where problems in docking may be isolated and understood. These systems are used to test new docking algorithms. The specific aims are: 1. To develop model experimental systems to test docking algorithms. We introduce seven model binding sites of increasing complexity. A. Five buried cavities in T4 lysozyme and cytochrome C peroxiadase: one purely hydrophobic, three bearing a single polar, cationic, and anionic residue, respectively, and a fifth that is anionic with several polar groups. These isolate polar, apolar, and ionic terms in docking scoring functions. B. A hydrophobic cavity in TEM-1, open to solvent at one end, that explores the influence of the protein-solvent interface in docking. C. AmpC B-lactamase, which has all the complexities of a drug-binding site, but is a system over which we have complete control. New docking methods will be tested experimentally against these model systems. Predicted geometries will be tested by crystallography and binding energies determined. In these systems, we expect to learn as much from "decoy" molecules that do not bind, as from those that do. These provide strong decoy sets for docking. Both the model systems and the decoys should be useful to the community. 2. To develop new docking algorithms. The docking problem is one of sampling available states, which grow exponentially with degrees of freedom, and evaluating their energies, difficult in condensed phases. To improve the energetic component in docking, we consider algorithms to A. calculate context-dependent ligand desolvation and B. receptor desolvation in docking. These key terms are often overlooked, owing to their high computational cost; the new methods should overcome this cost, albeit with important approximations. We also consider sampling C. receptor flexibility with an algorithm that scales only linearly with degrees of freedom, overcoming an otherwise exponential growth. This is extended to D. evaluate docking to homology models. All methods are tested experimentally against our model binding sites. Finally, we E. use the extensive sets of decoys developed in aim 1 to evaluate and improve docking algorithms.