Abstract
New distinct versions of known protein folds provide a powerful means of protein-function prediction that complements sequence and genomic context analysis. These structures do not supplant direct biochemical experiments, but are indispensable for the complete characterization of proteins.