We are developing automated techniques for identifying distantly-related members of protein superfamilies. Our work examines methods for automating the types of decisions made by biologists using intuitive pattern recognition, to remove some of the user interaction required. These ideas, combined with motif and cluster analysis, will enable us to generate superfamilies by examining large databases. This information is then interfaced to 3D structural information using the graphics visualization tool MidasPlus and others under development at the Computer Graphics Laboratory directed to this purpose. The combination with structural information provides additional links between distantly-related family members for which 3D visualization is essential. We are testing some of these tools to define distant sequence and structural relationships for human viral receptors and receptors associated with degenerative nervous system disorders.
Showing the most recent 10 out of 508 publications