Tools and concepts for Simulation Experiment Management in Systems Biology: Improving the production of simulation experiments through standard formats and management support.

Research goal.

In order to compare scientific results in the life sciences but also to integrate the outcomes from partners in large-scale research collaborations, standardization is necessary. The standardization and exchange of protocols by which data are generated are already widely promoted. However, for the same reasons that apply to wet-lab data generation, results from projects that involve mathematical modelling and computer simulations must be documented to improve exchange, reuse and reproducibility. For example, in order to reproduce a simulation plot in a publication it is often not sufficient to know the equations of the model. The choice of numerical algorithms and their internal settings can influence the simulation results. In parameter estimation the results are often driven by random number generation, requiring statistical information on the outcomes of parameter value optimization. Metainformation on simulation experiments improves the re-use of simulations, supports the development of new models from existing ones and helps reducing errors, thereby improving the reproducibility of scientific results in the field of systems biology.

In this project Dr. Dagmar Waltemath and her team investigate techniques for the encoding of simulation experiments. The project covers the standardized encoding of experiments in an XML format, supporting a range of types of simulation experiments, and including the versioning of both, simulation experiment descriptions and associated models. The exchange of simulation experiment descriptions, together with existing models, will help reduce the development time of models in systems biology, will help the reproducibility of publications and support training in systems biology.

Specific objectives.

• the further development of a standard for the description of simulation experiments (SED-ML). This will improve the reproducibility of model-derived results in publications.

• a study of model histories, tracing the development of a selection of models, describing the changes that have occurred. The analysis will help deriving criteria for the evolution of models, to enable reference to a particular model instance from a simulation description.

• a study of simulation experiment histories, tracing the development, describing and classifying the changes that have occurred. This will help to describe differences in simulation experiments.

• the provision of a simulation experiment management system for maintenance, public availability, retrieval, exchange, and versioning of simulation experiment descriptions in a standard format.

For further information and news please visit our project homepage.