|
We have been investigating ways in which the continuous process of protein folding can be broken down or segmented into discrete, combinatorial steps. Our aim is to be able to compare and classify multiple folding trajectories of the same protein (we have access to a lot of parallel trajectory data through the work of V. Pande at Stanford), or folding trajectories of similar proteins. The approach is based on the notion of looking at “proximities” that are formed as the protein folds between atoms that are not near each other in the molecular structure. The spanner structure described above is a natural tool for this effort, but it has to be augmented with better techniques for comparing spanners. As a by-product, our approach extends and unifies existing methods, including contact maps and distance matrices for structural similarity comparisons. The main aim of our effort, however, is comparison of trajectories and not of individual structures.