The research focus of this project is to develop a novel evolutionary computation-based approach for identifying candidate modules in non-coding DNA that respond to environmental toxins (such as arsenic) and that alter gene expression. These modules are composed of short pieces of DNA that are binding sites for proteins; the cooperative and combinatorial interactions are believed to contribute to the inducibility and specificity of environmentally responsive genes. Since each gene has an enormous number of possible modules, searching for them in the laboratory is untenable; even an exhaustive computational search for candidate modules is impractical, given the large space. Thus, the development of artificial intelligence techniques is called for.
This is an interdisciplinary proposal that makes contributions in both computer science and biology. The computational contributions include designing an effective search through the large and complex space of possible modules. While a few existing tools have been designed to search the thousand base pair region immediately upstream of the gene, the work here is designed to search significantly longer sections, 1 million base pairs and longer in length. The existing approaches cannot be expected to scale to the larger search, requiring the development of a novel approach.
The PI has plans for introducing undergraduates to research, both through coursework and in supervised research projects. This proposal will support and encourage the creation of a new upper-level course in informatics as well as the development of informatics-themed exercises to be incorporated at the introductory level. The project will further directly support undergraduate researchers who will contribute to the core research project. This project will result in a well integrated program of research and teaching for the PI, contribute to the available tools and our understanding of evolutionary computation approaches for informatics work, and introduce scores of students to this work.
