The distance geometry approach to conformational calculations has been very successful in producing conformations satisfying strictly geometric constraints, but there has been no way to weight the constraints or to include energetic considerations. I am developing an extension of distance geometry that not only handles geometric constraints, but also produces low energy conformations. Among the numerous applications for such a technique, I am particularly interested in studying protein folding, which is of course central to the basic understanding of molecular biology. My preliminary results indicate that the energy embedding extension to distance geometry deals with geometric constraints as successfully as always, and also produces conformers of very low energy. As the algorithm now stands, the calculations are quite time consuming, and a given protein under a given potential function will always come to the same final conformaton. Parts of the process can be speeded up by substituting faster computational techniques. Realizing that energy embedding is not guaranteed to find global energy minima, the algorithm must be modified for the sake of physical realism to produce several low-energy structures instead of just one. The choice of potential function must also be examined more thoroughly. Since the final structure depends so critically on the potential used, some functions may be better than others at guiding the molecule to a satisfactory final result. For predicting protein tertiary conformation, a particularly good """"""""potential"""""""" or set of constraints can be derived from the many empirical folding rules that are now known.
Showing the most recent 10 out of 15 publications
