Employing Bayesian Analysis in Pathway Modelling

Biochemical pathway modelling and analysis relies on the development of mechanistic models describing the kinetics of the main chemical species implicated in the overall function of the pathway. Such mechanistic models are highly parameterised representations of the biological system where the model parameters correspond to, for example, actual kinetic reaction rates. Such kinetic rates for biochemical reactions are mostly unknown, and only few can be identified using in vitro assays. Even for the rates that can be measured, biological systems tend to demonstrate a wide range of population variability. However the increasing availability of data obtained from various experimental protocols suggests the adoption of inferential methods in model identification and subsequent analysis. In particular it is argued that the Bayesian viewpoint is ideally suited for biochemical pathway modelling as issues surrounding system identification, sensitivity analysis, experimental design and model comparison are all straightforwardly addressed within this inferential framework. This talk will give an example of Bayesian analysis in modelling of the extracellular signal-regulated kinase (ERK) pathway where there are currently two working hypotheses as to how ERK is stimulated by extracellular growth factor (EGF) signalling. By modelling the pathway based on each hypothesis and employing the Bayesian inferential methodology we are able to objectively assess which hypothesis is best supported by the currently available experimental evidence.