Data-driven Inference of Cellular Networks
Genes and proteins of regulatory and signaling networks are often known whereas many of their mutual interactions remain still undiscovered or are unclear. Our group develops and applies novel algorithms for the computer-aided identification (inference) of cellular signaling and gene regulatory networks from experimental data. While we have used different formalisms for network inference (interaction graphs, logical networks, ODEs) recent work focuses on methods for interaction graphs:
- Sign consistency in interaction graphs as paradigm for network inference.
- Mixed-integer linear programming (MILP) and answer set programming (ASP) for solving combinatorial problems arising in sign-consistency based inference.
- Experimental design for discriminating candidate models in network inference.
- Transitive reduction in interaction graphs for identifying and removing edges from indirect effects.
We apply these methods for the inference of mammalian signaling networks, e.g. for the EGF and HGF signaling pathway in hepatocytes.