Joel Wagner (BE doctoral)
Our ability to measure protein levels and modifications in a high-throughput manner has outpaced our ability to assign kinetic rates to each of those proteins and their interactions with other species inside the cell. Differential equation models can describe the well-studied portions of the cellular network for which kinetic parameters can be reasonably estimated. However, improved experimental technologies have allowed more data to be collected for many more network components; consequently, modeling strategies beyond differential equations are now needed to describe the newly expanded but poorly understood networks.
In an effort to better capture the complexity of cellular networks by including many more measured species from high-throughput technologies—while still operating in a mathematically rigorous, quantitative framework—we propose the use of Bayesian networks. Bayesian networks are one method used for network inference, the process of determining the relationships among species that comprise the cellular network. Focusing on the epidermal growth factor and insulin signaling networks, we propose extensive use of the parameters of Bayesian networks (the conditional probability tables describing parent-child relationships), in addition to the structure of inferred Bayesian networks, to aid in therapeutic design.