Protein folding has important biomedical applications. The problem is under heavy scrutiny by decades of theory, simulation, and experiment. This paper reported the first full-atom simulation of a mini-protein in implicit solvent sampling enough folding events so the dynamics of the folding process could be compared quantitatively with experimental measurements of the kinetics.

The holy grail of folding modeling by molecular dynamics simulation is to get a protein to fold repeatedly in a full-atom representation, in agreement with experiment. This is an important step in that direction, merging new laser temperature jump technology and distributed computing technology into a coherent whole.

It's hard to compare experiments and computer simulations of how a protein folds into its three-dimensional shape: proteins usually fold too slowly for a computer to be able to simulate it. In our experiments, we studied an extremely fast folding mini-protein, and the computations were run on many thousands of PCs around the world to speed them up enormously, so experiment and theory finally met on a microsecond timescale in this Midwest-Westcoast collaboration.

By chatting about protein dynamics with a colleague of mine at Illinois who was working on the theoretical framework for folding. I became so intrigued by the idea of protein folding that I shut down one of my other research programs in 1994 and started working on this instead.