Exploring Protein-Ligand Interaction with Rotamer-Library Based Refinement Tools
Date of Award
Doctor of Philosophy (PhD)
Rotamer-library based techniques employing all-atom force fields have been extensively used for the predicting and refining protein sidechains, loops and tertiary structures. These refinement techniques have been expanded to allow the inclusion of protein-ligand interactions by developing and evaluating methods for creating rotamer libraries for small drug-like molecules. Two algorithms have been developed that use these rotamer libraries to refine a series of protein-ligand interaction poses. A hierarchical algorithm uses the flexibility encoded in the protein and ligand rotamer libraries to systematically choose and refine nearby conformations. This algorithm was used as part of an induced-fit binding protocol to refine and rank protein-ligand binding poses from apo and cross-docking test sets. The systematic optimization in this algorithm allows all-atom force fields to approach and in some cases exceed the accuracy of empirical scoring functions for protein-ligand interactions. A second algorithm uses a random-walk based approach consisting of random ligand and protein moves that are subsequently refined. This was used to find ligand diffusion pathways in cytochrome P450, fatty-acid binding protein, truncated hemoglobins and aryl alcohol oxidase. In the latter two cases, the pathways were identified as part of studies including quantum mechanical calculations and experimental results. The structure prediction methods, coupled with experiments, allowed the quantum methods to generate a complete picture of the biochemistry governing these processes even though atomic-detail data was only available for part of it.
Tao Ju, Garland Marshall, Jay Ponder