---------------------------------------------------------------------
    COLLOQUIUM OF THE LABORATORY FOR COMPUTER DESIGN OF MATERIALS
        Institute for Computational Sciences and Informatics
	                      (CSI 898-Sec 001) 


              Ab Initio Protein Structure Prediction and a 
                      Comparative Genetic Algorithm

                          Susan K. Gregurick
           Center for Advanced Research in Biotechnology, 
        Univ. of Maryland Biotechnology Institute, Rockville, MD

In the present study we address an important problem in ab initio 
protein structure prediction, which is how to avoid defects in the 
potential energy surface that ultimately lead to incorrect structures.  
These defects correspond to local minima and are hard to avoid in most 
simulations.  The resulting algorithm, loosely termed a Comparative 
Genetic Algorithm, is similar in spirit to the idea that multiple energy
surfaces can be explored simultaneously by using homologous sequences.  
This arises from the observation that homologous proteins often share 
the same global minimum but do not share the same incorrect local minima.  
In the Comparative genetic algorithm multiple potential surfaces, defined 
by different potential energy functions, are searched simultaneously 
by optimizing two different energy objective functions.  This results 
in an algorithm which will avoid incorrect non-coincident local minima 
during the searching procedure.  It will be illustrated that in many 
cases we were able to substantially enhance the likelihood of finding 
a correctly folded structure.

	                  Monday , March 20, 2000
                                  4:30 pm
                         Room 206, Science & Tech. I

Refreshments will be served.
----------------------------------------------------------------------
Find the schedule at http://www.csi.gmu.edu/lcdm/seminar/schedule.html