Navigating high-throughput docking results

CINF 77

Keana Scott, Noel Southall, Trung Nguyen, and Dr Ajay. Informatics, Celera Genomics, 45 W. Gude Drive, Rockville, MD 20850
High-throughput docking results are often subjected to strict filters that are based on multiple scoring functions and applied in linear fashion to reduce the docked poses to a manageable number for visual inspection. Although filters reduce the number of false positives, they also decrease statistical power in the docking exercise by increasing the number of false negatives. Instead, we have developed a flexible tool that 1) allows the user to navigate through the entire binding mode hypothesis space rather than a small subset of individual poses, 2) does not preclude interesting unanticipated binding modes, 3) incorporates multiple in-house developed scoring functions, and 4) enables us to leverage a modelerís intuition. This flexibility is achieved through a Java-based user interface that allows for mathematical/logical operations in hypothesis space and works hand-in-hand with PyMol for visualization and an Oracle backend for data storage.