Title:  Protein Folding: A Global Optimization Problem


			   Ken A. Dill
			 UC San Francisco
			  600 16th Street
		      San Francisco, CA 94158
               Email: dill@maxwell.compbio.ucsf.edu
                 Web page: http://www.dillgroup.ed


Abstract: 
A protein is a molecule with tens of thousands of degrees of
freedom.  It folds to a unique structure (a global minimum in the free
energy function), on a complex rugged landscape, often in microseconds to
miiliseconds.  It is clear that the protein cannot do a globally exhaustive
search to find this state.  We find that the protein has a two-part
strategy.  First, the landscape has a general funnel shape, which speeds up
the early stages of the search.  Second, we believe proteins use a strategy
we call 'zippers', in which the global optimum is found by a series of local
optimizations.  Based on these observations, we are developing an algorithm
that might be useful in protein structure prediction.