We expect that the theoretical foundations laid down in this project will provide tools as well as a rational basis for hierarchical composition of multi-level models in Computational Systems Biology. At present, there is no general, logically sound way to combine biological network models, or to decompose a model into more or less autonomous sub-units. Problems encountered in model composition vary from purely practical to rather fundamental: there may be conflicting use of identifiers, incompatible dynamical descriptions, differing abstraction levels or scenarios, inconsistent hierarchical compartmentalisation, omission of input and output channels, and numerous other issues. In model decomposition a similar set of problems is encountered, with prominent ones relating to the incompleteness of models, and the abundance of feedback loops that prevent the model to be organised as acyclic graphs.

Although attempts to specify the elements that should comprise a model’s interface have been undertaken, progress has been limited, and no agreement has been reached so far within the communities that attempt to tackle the standardisation of biochemical network modelling (e.g. the SBML, BioPAX, and CellML consortia). Because Krohn-Rhodes analysis reveals the hierarchical algebraic structure of models, we argue that this hierarchy will provide a rational basis for the partitioning higher-level models into a set of lower-level ones. These lower-level models may then be expanded (“manually”) to include greater detail, and again subjected to Krohn-Rhodes analysis. BIOMICS will therefore develop new and extend existing computational systems biology modelling frameworks based on a non-linear systems perspective and its application to the software systems develop