Parameter estimation has become a central problem in systems biology, both in the form of calibration of bottom-up models or as a component of reverse engineering algorithms. With a proliferation of algorithms proposed for these purposes it has become important to compare them in objective ways. I will argue that in silico biochemical network models are extremely useful for this purpose. Several networks will be presented that are challenging tests for parameter estimation and network inference. An issue that arises from the use of in silico networks, though, is whether they can provide realistic data. The application of this benchmarking methodology will be illustrated with a comparison of four reverse engineering methods.

Joint work with Diogo Camacho, Paola Vera Licona, and Reinhard Laubenbacher