Characterization of systems-level organization through mining large-scale genetic interaction networks

Understanding cellular organization and function on a global scale is a central challenge in systems biology. Recent high-throughput experimental technology has produced datasets capturing physical and genetic interactions among various components of the cell, which contain clues about systems-level organization. While the analysis of physical interaction networks has received significant attention from the bioinformatics community, relatively little attention has been devoted to analysis of the more general phenomenon of genetic interaction. Genetic interactions are generally defined as cases where simultaneous mutations in multiple genes result in a surprising phenotype, and are highly complementary to information in the physical interaction network. In collaboration with a yeast genetics lab, we have developed a strategy for highly quantitative characterization of interactions from synthetic genetic arrays (SGA), and are now getting our first view of a large fraction of the yeast genetic interaction network. We will introduce several fundamental concepts supporting the emerging direction of genetic interaction networks and describe several striking properties revealed by mining the structure of such networks. For example, we will discuss the connection between modularity and genetic interactions, and describe the potential of these data for global characterization of network structure and robustness. We will highlight several open problems in the interpretation of such networks and discuss where innovations in machine learning and data mining are particularly relevant.