Protein Modeling in the Structural Genomics Era

Donald Petrey

Columbia University

The goal of the structural genomics (SG) initiatives is to experimentally determine the structures of enough proteins to allow the use of computational or other techniques to determine the structure or function of all proteins.  An important part of this effort is the identification and quantification of the similarity between protein structures.  I will discuss this question and techniques we have developed to identify structural similarity as well our attempts to use this information to infer function.  I will also discuss alternatives to existing protein structure classification schemes that are more appropriate to the protein structure/function prediction tasks generated by the SG efforts.  The way we define and represent structural similarity between proteins also has an effect on the protein structure modeling problem, and I will discuss our recent work in using new definitions of similarity to develop sequence-to-structure alignment and model building algorithms.