The genomics revolution has uncovered tens of thousands of proteins, nonetheless full exploitation of organismal genome projects will not occur without innovations in proteomics and structural biology. The agencies meeting this challenge, such as the seven NIH protein structure consortia , have targeted 10,000 new protein structures to be solved by 2010. We propose to use this powerfully enabling structural dataset as the basis for a new type of bioinformatic tool which models and compares the surfaces of proteins, instead of their sequence or backbone path, by analogy to sequence comparison methods. An algorithm that searches a protein's sw-lace features for post-translational modification and binding sites will offer a large step forward. The ability to model protein binding sites will allow investigators to predict how they interact, how regulatory networks are organized, and what determines the order of metabolic pathways--each a fundamental goal of much research. Cognia is well positioned to undertake this effort as we are currently building databases and tools designed to identify drug targets and biotherapeutic candidates for use in both basic research and pharmaceutical-biotech discovery. Our use of predictive tools to analyze, organize, and synthesize the content of databases lends direction to this proposal, and its near-term focus will involve the analysis of post-translational modifications such as the ubiquitylation and SUMOylation accomplished by E3 enzymes during the cell cycle. We anticipate that algorithms developed under this proposal will have many applications, including the ability to (I) predict novel binding interactions of structurally defined proteins (ii) identify structural homologs of proteins not apparent by sequence or backbone similarity searches (iii) define an enzyme's potential substrates, (iv) define structural features of binding sites. In Phase I, we will gather the requisite data and structures and carry out a pilot definition of an E3's consensus binding site. This will provide the foundation for Phase II, where we will create a tool that will scan all known structures for consensi, predicting interactions and modifications.
