Pitt faculty receive awards to explore next-generation technologies

Di Gao could help usher in the much-heralded future of personalized medical care based on an individual’s DNA in his effort to revamp the technique for screening and separating DNA molecules. Gao’s approach would stretch DNA strands tethered to a solid surface via an electric field, allowing them to be pulled from the surface and analyzed based on their viscoelasticity. This method would overcome the limitations of the predominant method of electrophoreses—submerging the strands in a matrix and applying an electric field. By stretching the DNA, chromosome-sized DNA molecules can be separated and studied, large fragments can be screened for mutations, and longer sequence fragments can be extracted. The technique might also be applied to RNA. The education component of Gao’s project includes outreach to minority high school students through a related course and workshop at Baldwin and Westinghouse high schools in Pittsburgh, both of which have large African American student populations, and collaboration with Tsinghua University in China on an international field study module for Pitt undergraduates that focuses on international views of the ethical and social issues of genetic research.

Rebecca Hwa aims to improve the ability of computers to process and translate human language. She will address the difficulty many systems have in processing texts from such specialized domains as business emails or scientific literature as well as texts that are automatically translated from foreign languages. Specifically, Hwa will create machine-learning algorithms that find correspondences between “standard English” and texts from specialized domains. The project focuses on three types of correspondences: direct translations, such as bilingual documents; loose translations, e.g., paraphrased articles; and indirectly related texts without an explicit translation. Building from these correspondences, a standard system will be adapted to translate texts in specialized domains. Better language processing for a wide range of texts could allow for such computer applications as intelligent tutoring programs and data mining for medical documents.