An efficient and effective conformational search method is essential for solving the computational protein folding problem. Prevalent conformational search models are fast but incomplete. Proposed here is an exhaustive enumeration-based search method that combines a focus on the most promising search paths with a complete coverage of the conformational space. The search is a prioritized building-up process: Structural templates of low energy are generated for short segments of the chain. Structural elements, either templates or individual residues, are added to a growing conformation one at a time. The addition of a structural element is prioritized on its energetical contribution one at a time. The addition of structural element is prioritized on its energetical contribution to the conformation as a whole. Simple templates such as tight turns, helices or strands form packing clusters such as hairpins and beta-sheets. A finished conformation is often the assembly of packing clusters. Such efficiency is achieved by pruning provably unpromising search branches at each building-up step. Phase I will implement the architecture of prioritized assembly-based search, demonstrating the feasibility of the approach. Phase II will develop algorithms for computing tight lower bounds of conformational energy, which are crucial for effective branch-and-bound search. Phase III will develop a comprehensive conformational search program for computational protein folding:
The primary market will be the academic research labs and technology concerns that need diverse sets of possible conformations for investigating peptides, proteins and protein-protein complexes. An additional market is the research labs that need near native or misfolded protein structures (decoys) for testing and improving their energy potentials and search methods.
