Beginning last summer, we embarked on the study of algorithms related to the folding of globular protein molecules. Dr. Kang started by investigating the feasibility of making local dihedral angle motions. We found that perfectly local motions are not feasible but that motion that preserves global orientation and introduces only small global translation are possible. We have now written a computer program that incorporates these local motions in the Monte Carlo scheme in the dihedral angle space. The development of this program is progressing rapidly. It can generate protein structures according to sequential or multinucleated mechanisms. It does dihedral angle space dynamic Monte Carlo using predefined probabilities for the dihedral angles learned from scanning many known structures. At present, only the hydrophobic force drives the system. We feel that this force is the most important for folding, but a full, successful operation requires including other, more subtle forces including the hydrogen bonding, the van der Waals force, and electrostatic interactions. These forces will be added to our programs in the future.
