Model and simulation annotation

TR&D 2 (annotation) aims to accelerate biomodeling through enhanced annotation of models, simulation experiments, and simulation results. As models accumulate in public repositories, there is an opportunity to reuse models for new studies and to combine models into comprehensive meta-models of entire biological systems. However, it is currently challenging to reuse and combine models because few models are reproducible or understandable. Consequently, modelers currently waste huge amounts of time trying to understand, reproduce, and combine models published by other modelers, including other modelers in the same research group. To make it easier to understand, reproduce, and combine models, we must make the assumptions, meanings, and limitations of models explicit.

We have begun work toward these goals by delivering a set of standards and technologies around semantic annotation. First, we have collaboratively developed the OMEX Metadata Standard, so that best practices and expectations for how to annotate models are established and agreed on by the modeling community.

Next, we have delivered a set of software tools that use this standard to enable users to appropriately annotate models, and then use these annotations to retrieve appropriate models. Details are provided below under software.

To ensure these tools accelerate biomodeling, this TR&D includes a number of Collaborative Projects which need enhanced annotation schemas and tools to understand, reproduce, reuse, and merge their models. These Collaborative Projects will push us to develop user-friendly graphical interfaces to our tools, and we will pull the Collaborative Projects to use our new annotation standards and software.

We have also established a critical Service project that provides curation capabilities. Working with PloS Computational Biology on Improving reproducibility in computational biology research, we have established a pipeline whereby authors submitting papers with models, can, as part of the peer review process, validate whether or not their model is reproducible. We produce a reproducibility report, and this can then be used to promote and better share reproducible models in publications.

Moving forward, we will continue to expand our curation work, and develop additional annotations for model credibility. In addition, we are exploring new ways to merge and connect models across different modeling paradigms. The methods, tools, and services provided by this TR&D will help modelers discover models for new studies, better understand published models, and augment and merge models to test new hypotheses about physiology and pathophysiology.