Systematic discovery of condition specific Wnt signaling subnetworks

Wnt signaling is a critically important signaling pathway in development, differentiation and embryogenesis and is broadly conserved amongst multicellular animals. Dis-regulation of Wnt signaling is associated with many human cancers, in particular colorectal cancer. Multiple branches of the Wnt signaling pathway (`canonical´ and `non-canonical´) have been defined, although increasingly these are viewed as a single interconnected signaling network. A principal challenge in Wnt signaling research is to understand the connectivity between different branches, and to understand how different branches of Wnt signaling function in different biological states. To better understand this complex network, we analyze Wnt signaling in the context of diverse cell-lines, tissues and adenomas. By integrating gene-expression datasets representing these different biological states with interaction networks, we identify Wnt subnetworks activated in different biological states. We identified 10 active subnetworks in the network of 2072 protein-protein interactions from a Wnt-focused interaction network. These subnetworks were used for classification of samples and we show that classification of these samples using subnetworks rather than gene-expression profiles alone identifies potentially significant commonalities between samples. Specifically, we identify a subnetwork common to lung and colon adenoma samples, reflecting the common origin of these tumors. The approach described here will be used in future for integration of diverse Wnt-focused `omics´ datasets, for both defining components of the signaling network itself, as well as to reveal commonalities and distinctions between Wnt signaling under different biological states.