Abstract

Similarities between different protein structures have led to the identification of protein families based upon some measure of structural similarity. Using these similarities one can classify proteins into structural families and higher-order groupings from which inferred function can be transferred. When taken for a large number of proteins, these schemes point to evolutionary relationships between organisms. We propose a novel classification scheme based upon the structurally-inspired dynamics of each protein. This classification scheme has the advantages of being quantitative, automatically assigned, and able to also make distinctions within protein families. Results are presented for five protein families illustrating the correct identification of previously un-classified structures and sources of intrafamily distinctions.