Quantitative model refinement

We focus in this project on computational techniques allowing mathematical models in biology to be presented and used at different levels of detail. We aim to construct and investigate a computational framework for quantitative model refinement where details may be added to (or removed from) a model in a systematic, semi-automatic way, while preserving its numerical behavior.

Our methodology addresses a major current challenge in computational systems biology, that of specifying a model at various levels of resolution while formally ensuring that once the model is quantitatively fit and validated at some level, it remains so at any other level. In this way, a model can be specified first at a lower level of detail, so that its numerical fit is computationally efficient. Once the model is fit and validated it can then be formally refined to higher levels of detail based on our framework. This approach provides the basis for creating and handling flexible hierarchical models, able to integrate data and processes across scales. The project is of a methodological nature, focusing on the development of a sound computational framework for quantitative model refinement. Its tools and methods belong mainly to theoretical computer science. The main contribution of the project will be to computer science by establishing a framework for the quantitative refinement of computational models, a topic not yet investigated much in computer science. This framework is a quantitative development of work done in connection to program specification and (qualitative) refinement, at the borderline between computer science and software engineering. The project will also contribute to computational systems biology as an application area of our methodology, by creating a sound basis for multi-resolution modeling of biochemical systems. We test the applicability of our methodology on several biomodeling case studies.

Throughout our project we focus on the following three objectives:

Construct algorithmic foundations for quantitative model refinement;
Construct a framework for quantitative model comparison;
Demonstrate the applicability of quantitative model refinement in systems biology.