SHARPEN and Folding@Home
- Contrasting structure prediction with molecular dynamics
- Native structure discrimination
- Combinatorial optimization
- Stanford - Colorado State University collaboration
A dominant theme for Folding@Home has been molecular dynamics simulation, simulating Nature on the molecular level, femtosecond by femtosecond. The focus of SHARPEN is different. Rather than simulating the process of protein folding, we attempt to efficiently jump to the best protein structure.
One of the crucial elements of a structure prediction algorithm is the energy function used to discriminate between the best structures and the astronomical number of similar alternatives. We are implementing molecular mechanical energy functions as well as energy functions that mix MM terms with knowledged-based (statistical) terms such as Rosetta.
Many algorithms have been used for rotamer optimization including Monte Carlo (simulated annealing), Dead End Elimination, FASTER, Mean Field, Dynamic Programming, and Graph Theoretical approaches.
Our site at CSU hosts the server responsible for the SHARPEN Work Units. The Folding@Home team at Stanford runs the Assignment server that hands out assignments to the volunteer clients.