Dimopoulos, Sotiris

For dynamical systems, such as signal transduction pathways, our current abilities for development of predictive mechanistic models is rather limited. Either due to limited availability of experimental data, or due to unknown components and their interactions, large uncertainties in both model structures and parameter values occur.

In my project, I plan to address the question of how one can systematically represent uncertainty in biological mechanisms via ensembles of detailed mathematical models. To do so, probabilistic frameworks that quantify the significance of dynamic model differences need to be developed. These frameworks have to take into account the uncertainty in model behavior, the model likelihood and the uncertainty of experimental observations. Thus, it is essential first to develop core mechanistic dynamic models that will serve as an initial model ensemble to the framework. Using yeast as an example organism, I currently perform targeted quantitative time-course experiments that aim to extend existing dynamic models of glucose signaling pathways, mainly towards nutrient sensing mechanisms. The acquired experimental data are also taylored to serve for model discrimination and validation. Therefore, a tight integration of experimental analysis and mathematical modeling is performed. Through this approach, not only model construction and analysis profit from available measured data, but the model, in turn, can be employed to stimulate and accompany the design of experiments and facilitate interpretation of results.

This work is carried out in the groups of Jörg Stelling and Matthias Peter .