Wednesday, March 16, 2016 4:00pm - 5:00pm
Watson CIT - SWIG Boardroom (CIT241)
Mining Gummi Bears in One and Two Jars
Integrative network analysis methods provide robust interpretations of high-throughput molecular profile measurements in the context of gene networks. Active subnetwork modules are sets of genes that are responsible for observed phenotypic changes. In our model, finding the most active subnetwork is equivalent to finding a maximum weight connected subgraph (MWCS) with respect to appropriate gene scores. MWCS is NP-hard. In this talk, I will describe Heinz, an exact approach to this problem based on graph-decomposition and integer linear programming, and eXamine, a Cytoscape app to explore the resulting modules. I will illustrate the methods by means of an application to virally deregulated signaling.
I will also report about recent developments in the xHeinz project, which explores novel concepts to find transferable modules between model organism and human. Transferable modules are both active and conserved. The model therefore needs to take aspects of comparative network analysis into account.
Hosted by Ben Raphael