Development of a facile, web-based environment for the characterization and detailed comparison of protein surfaces is proposed. Five principal components will be utilized: 1. a web server (providing surface viewing capabilities to non-expert users); 2 a database (consisting of processed PDB files); 3. a suite of comparison tools ; 4. a detailed characterization of proteins within selected families, and 5. a novel query engine. Given the fact that important features can differ significantly in prominence, it has proven difficult to study protein surfaces for recognition and comparison. To deal with this difficulty, the investigators have adopted a hierarchical topographical primary sketch (TPS) from computer vision that permits varying degrees of surface smoothing. Features extracted from this hierarchy are used to 'tokenize' the surface. The tokens so provided correspond to features that can be naturally ranked by topographical prominence. The TPS organize a database that contains a full biochemical/biophysical description of underlying surface patches. This enables a rich specification of descriptors that can be used for matching in database queries and allows for meaningful comparison of surfaces within protein families. The program GRASP will be used to extend the surface visualization capabilities, and coupled to the server's query engines and underlying databases. Thus, the user will be provided with a timely tool for the interpretation and comparison of structure-function relationships within and across protein families. This search engine will be important to the emerging field of structural genomics.
NOT AVAILABLE
