Metamodel and Modeling Language: towards an Unified Modeling Language (uml) Profile for Systems Biology

Since data integration is a first-order priority in systems biology, metamodeling that is the foundation for data integration, should become an effective strategy among current approaches. In practice, the building of a metamodel requires four levels: the information level 1 or data level, which consists in the basic facts to integrate, the data model at level 2, i. e. how the data are organized, the metamodel (level 3) that describes and organizes concepts with a set of well-formed rules to integrate models from level 2 and at level 4, the language for metamodeling that uses concepts and relations defined in the metamodel. Based on these principles, we developed an object-oriented systemic metamodel understandable by humans and computers that makes it possible to integrate data on structure, on the one hand and function, on the other hand. This metamodel was grounded on the active class, FunctionalUnit, whose attributes specify time-, spaceand shape-dependency of biological functions. A special emphasize was put on the “shape” attribute, named FormOccurrence, to account for the multifarious functional isoforms with respect to the gene or the protein from which they are derived. The context-dependency of biological functions was assessed through specific roles devoted to the FunctionalUnit class, they are vicinity (NeighborFunctionalUnit) for local context-dependency and globality (SupraFunctional Unit) for global contextdependency. We used the Unified Modeling Language (UML) to design our metamodel and we present arguments for customizing it to systems biology.