Identification of active biological networks and common expression conditions

Biological networks such as protein-protein interaction networks and metabolic pathways are useful in understanding biological processes within living cells. The fact that some parts of the networks are activated under specific conditions, while other parts work continuously, prompts us to determine the parts and conditions. In this paper, we present a novel computational approach to simultaneously identify the locations and the conditions from gene expressions and protein interaction data. Our approach is based on the determination of large interaction networks whose genes share expression or repression conditions. We evaluate our method by using yeast protein-protein interactions and microarray expression data. The experimental results show that our method can extract networks associated with specific conditions; these networks are closely related to the Gene Ontology categories or KEGG pathways. Our method identifies the components in a proteasome complex along with their activating conditions and the relation between heat shock and cytoskeleton automatically.