With all the different projects Professor Chi-Ren Shyu has on his proverbial plate, it's hard to imagine he has any time to sleep. Yet with easy finesse and exuberance, Shyu describes just a few of his ongoing "joyful and rewarding" research initiatives, ranging from biomedical and geospatial informatics to computer imaging of medical images. Not surprisingly, Shyu has gained a well-earned reputation for his collaborative work. Although diverse, what these research interests share is the effort to create large-scale, fast, and multidimensional databases.
For instance, one of his team's projects involves developing a "protein structure retrieval system" (a.k.a. protein database), with which the user can upload a 3-dimensional protein structure to the database to search for a match. Irreverently referring to it as a "Google for protein structures," he notes that this new database is the first of its kind in the world that operates in "real time"--that accomplishes the search in seconds (versus the days or weeks required by previous tools). This instrument could be used, for example, in developing medicines that target particular proteins, in addition to multiple other applications. In fact, as soon as this exciting innovation was made public via an article in Science magazine in September 2004, numerous people from around the world started to access the system (free of charge) at http://proteindbs.rnet.missouri.edu.
A second project, supported by the National Geospatial Intelligence Agency (NGA), one of the three United States intelligence agencies, involves developing a geospatial database system for analysts to quickly relate new images to an existing archive of maps and satellite photographs. The present system requires the analyst to manually pull out map books in order to compare images, a time-consuming process. With Shyu's new search engine, called "GeoI," the process will be complete in mere seconds. For instance, the agency personnel will be able to submit a satellite image to the database, highlighting the area of interest (e.g., an area under construction with intersecting roads and buildings), and search the world for similar sites. While, because of security reasons, this particular system will be accessible only to the U.S. government, it is likely that the team will eventually develop a version for commercial uses.
Supported by the National Science Foundation under the faculty early career development program (CAREER), the third project that began in July of this year seeks to understand and classify the physical appearance or phenotypes of plants. Different mutations and diseases in plants produce different appearances; in corn, for example, these phenomena might affect the color of kernels, the texture of leaves, or the type of root. Shyu's team is developing a search engine to which researchers (or educators and agriculturalists) may submit a photograph of the plant, a free text of gene description, and a DNA sequence of the plant in question in order to study possible mutations or diseases. The current method requires the agricultural specialist to painstakingly compare the mutated or diseased plant with images in textbooks, a process that is frustratingly slow, limited in the number of images referenced, and often hampered by outdated information. Shyu and his collaborators launched the first prototype for the plant research database in January 2006. Like the protein database described above, this "Google for plant phenotypes" will also be available to the public free of charge.