Challenges in predicting ligand binding free energies in a model binding site
Feb 22, Thu 2007
3:00pm - 4:00pm, 2001 Malott
Dr. David Mobley
The University of California San Francisco
A central challenge in structure-based ligand design is the accurate prediction of binding free energies. I will discuss my work applying alchemical free energy calculations to a model hydrophobic binding site in T4 lysozyme that has proven a challenging test system for docking methods. Retrospectively, these methods were able to compute absolute binding free energies within 1.9 kcal/mol of previously determined experimental values on average. In prospective tests, the methods correctly discriminated between several true ligands and decoys from a set of putative binders identified by docking. The X-ray crystal structures of the ligands were determined and compared to both the alchemical and docking predictions. We also quantitatively predicted and confirmed binding free energies for several compounds. Even for this simple model binding site, there are several challenges, including (a) the presence of multiple ligand orientations and (b) frustrated protein sidechain degrees of freedom. Resolving sampling problems associated with these challenges leads to improved agreement with experiment, and highlights the importance of conformational heterogeneity in ligand binding.