Connecting Models to Data in Multiscale Multicellular Tissue Simulations

AbstractSystem level biological behaviour typically arises from highly dynamic, strongly nonlinear, tightly coupled interactions between component processes occurring across multiple space and time scales. The interdependent nature of these processes often makes it difficult to apply standard mathematical techniques to separate out the scales, uncouple the physical processes or average over contributions from discrete components. To make rapid progress we need to address interoperability challenges: to build integrated models from reusable components, and to relate simulation results to experimental data both for parameter fitting and model analysis. In this paper we describe how work we have done to address these issues in the domain of cardiac electrophysiology can be applied in a completely different field: multicellular models of intestinal crypts, with cells treated as discrete entities, and the sub-cellular, cellular, and tissue scales interacting. In this application the model and simulation are intertwined in software, with no suitable markup language model representation. Different modelling paradigms are available for each of the scales, and comparing their predictions is of particular interest. We use our concept of ‘functional curation’ to separate the experimental protocols applied to models from the model descriptions themselves, allowing easier comparison of model behaviour with experimental data. We also describe the use of ontological annotation for providing semantically rich model interfaces, facilitating coupling models to each other and to protocol descriptions. Finally, we show how these uses of semantic annotation and markup languages may be mixed incrementally with legacy code. This work suggests that the ideas we have developed have the potential to be useful across computational science, and we discuss these wider implications