Quantitative Biological Models as Dynamic, User-Generated Online Content

Abstract

Quantitative models of biological systems are dynamic entities, with modelers continually tweaking parameters and equations as new data becomes available. This dynamic quality is not well represented in the current framework of peerreviewed publications, as modelers are required to take a 'onedimensional snapshot' of their model in a particular state of development. The CellML team is developing new software for the CellML Model Repository that treats models as dynamic code, by providing serverside storage of models within a version control system, rather than as static entities within a relational database. Version control systems are designed to allow a community to collaborate on the same code in parallel and to provide comprehensive revision histories for every file within the repository. Because CellML 1.1 espouses a modular architecture for model description, it is possible to create a library of modular components corresponding to particular biological processes or entities, which can then be combined to create more complex models. For this to be possible, each component needs to be annotated with a set of metadata that describes its provenance, a detailed revision history and semantic information from ontologies. A high level of collaboration is also essential to leverage domainspecific knowledge about the components of a large model from a number of researchers. By treating quantitative biological models as dynamic, usergenerated content, and providing facilities for the expert modeling community to participate in the creation and curation of a free, openaccess model repository, the the nextgeneration CellML Model Repository will provide a powerful tool for collaboration and will revolutionise the way cellular models are developed, used, reused and integrated into larger systems.