A variety of molecular docking applications can be cast as optimization problems, where achieving a minimal energy state corresponds to a physically interesting docking configuration. However, a simple application of optimization techniques may not provide adequate insight into the underlying molecular docking application. We describe two such applications: flexible docking of ligands in AutoDock, and a computational phage display technique for prediction of protein-protein interactions. A common element in these applications is the need to broadly understand the set of near-optimal solutions. This has motivated our development of COLIN, an optimization middle-ware framework that supports the caching of solutions for post-optimization analysis. COLIN is particularly well-suited for complex, hybrid optimization techniques, since it can manage caching of solutions from different optimizers that employ different problem representations.
Back to Workshop II: Optimization, Search and Graph-Theoretical Algorithms for Chemical Compound Space