9 - Data without models merging with models without data

Abstract

Systems biological models aim at a detailed account of the dynamics of complex biological systems, where “detailed” means that not only basic qualitative characteristics of the system but also more specific properties of a dynamical network are reproduced by the model; in particular, the ontology of the model should refer to the systems' physical components. Such models can be separated into two classes with different historical roots. The two classes of models are often characterized as bottom-up and top-down, respectively. The idea behind this nomenclature is that some models are built from data concerning well-characterized components of the system—the “bottom-level” of the biological system—so that the system—or top-level is reassembled mathematically from the components' contributions, whereas other models are built from data concerning the system as a whole and try to break down the system mathematically into modules and components. Bottom-up models typically consider data on a few (less than ten) components only, while top-down models may be based on data about as many as hundreds or thousands of components. Therefore, bottom-up models are classified as data-poor models, while top-down models are considered as data rich. A similar difference in data richness is characteristic of the forerunner approaches on which systems biology is based upon, although in this case the borderline is not between bottom-up and top-town but between modeling and the acquisition and analysis of complete data sets with respect to a particular domain, such as in genomics.