Generating plausible protein folds by secondary structure similarity

We extend a standard method of comparing protein sequences to develop a method for quantitatively comparing the secondary structures of two proteins and finding the best alignment between them. We introduce a new approach to generating a set of plausible folds for a protein given only an assigned secondary structure (a query) and use this secondary structure comparison method to search the set of proteins with known tertiary structures for an array of secondary structure elements similar to the query. A large fraction of the similar arrays found in known proteins have folds that can be taken as candidate folds for the query. The approach is analogous to “protein extension” based on primary structure. We show, for example, how the method can produce plausible folds for the secondary structure of flavodoxin, including the correct one.