Inference in systems biology: modelling approaches and applications

The main topic of this thesis is the study of biological regulatory systems using different computational modelling approaches in order to gain new insights into not yet completely understood biological processes. In "systems biology", mathematical models represent a powerful tool to study biological processes. Models are abstractions of reality always including some degree of simplification: an important ingredient of the modelling process, having a major role in suggesting the appropriate level of abstraction and simplification, is the purpose of the model, that is the question they have to answer. This thesis is focused on the analysis of how models of different complexity appropriately describe the available data to achieve a given purpose. Such analysis guides the choice of the most appropriate degree of simplification of the system under study that allows neglecting some aspects without compromising the results of the model. Three levels of detail for inference and modelling are analyzed in this thesis depending on the system under consideration. The first level is the network level, where molecules are nodes connected by edges and the interest is in the inference of the topology of connections at large scale. In the second level the network is interpreted as a mean to produce qualitative simulations and predictions which can be compared with experimental data. The third level of detail consist in a more mechanistic dynamic description of the system using ordinary differential equations but limiting the analysis to small subsystems. For each level of detail, appropriate approaches have been developed and applied to in silico and real data of different biological systems. Finally, different modelling appraches have been integrated to analyze insulin signalling pathway on different levels of simplification using a novel experimental dataset collected specifically for this purpose