Determining System Boundaries in Yeast using Genetic Interactions and Known Pathways

The eukaryotic cell is often described as a hierarchy of systems (complexes and pathways) composed of genes or proteins. These systems can be defined at multiple levels (e.g. one complex containing several subunits or different complexes interacting together). Identifying these systems and their relationships from experimental data is challenging. Hierarchical clustering [1] or network analysis [2] has been proposed to extract modules from genetic interaction networks. These methods are useful but define a flat set of modules often describing complexes. Boundaries and system level relationships are still difficult to define. Strategy We propose to use known pathways, complexes and their relationships as defined in Gene Ontology (GO) annotations to better define cellular system organization. We assess the coherence of systems defined by the biological process ontology using quantitative genetic interactions [3] which are roughly classified into positive and negative interactions. It is known that, often, positive interactions are found within systems and negative between systems [4]. We use this idea and characterize a gene set by it