November 2008 - October 2010
Department of Clinical and Experimental Medicine
SE-581 83 Linköping
Systems biology of insulin signalling by unique identiﬁcation in overparametrised models
Insulin signalling is at the heart of glucose homeostasis and its malfunctioning in type II Diabetes, a rapidly growing disease of world-wide proportions. For this reason insulin signalling is extensively studied. However, the complexity of the signalling protein network has so far greatly limited the progress; the complexity makes it impossible for classical biochemical reasoning to be the only source of data analysis. For this reason, I am working with systems biology approaches, which use mathematical modelsforthedata analysis.
I am currently building up my own systems biology group, as a subgroup to an experimental insulin signalling group working with adipocytes. I therefore both have a general background in some of the general theoretical problems, as well as with details of insulin signalling. One of the most important of the general problems in mathematical modelling is over-parametrisation. This occurs because signalling systems are large, and because current measurements are semi-quantitative and only cover a few of these proteins. This means that almost all model predictions are non-unique, and hard to make us of. During the last 7 years I have developed various methods to overcome this problem. In particular I have developed the concept of a core prediction, which is deﬁned as a uniquely identiﬁed model prediction, even for an overparametrised model. Such corepredictions are of course highly interesting conclusions from a data-analysis, and I have developed several methods for determinations of such predictions. Some recent and related methods have also been developed in the group of Jens Timmer , who also is working on insulin signalling, but in hepatocytes. I will utilise some of these related methods to develop improved methods for identiﬁcation of core predictions. The developed methods will continuously be tested on the data analysis within the insulin signalling studies, inparticular to draw conclusions regarding feedbacks and time-scales of the insulin receptor dynamics in single cells. Although the stay is only limited to one month, it should be emphasised that this will be suﬃcient to get some preliminary results, which will allow for a continued collaboration also after the stay is over.
G.Cedersundet al. Modelbasedhypothesis testingofkey mechanismsininitialphase ofinsulin signaling. PLoS Comp Biol 4, e1000096, 2008
 S. Hengl et al. Data-based identiﬁability analysis of non-linear dynamical models. Bioinfor-matics 23, 2007, 2612-2618