Reproducible COVID-Models

European Open Science Cloud

The onset of the COVID-19 global crisis demanded fast support for rational predictions of epidemiological systems behaviour to inform potential interventions such as therapeutic measures, and for education and information dissemination such as explaining what is “flattening the curve”. Virtual studies about COVID-19 to the scientific community such as this one and this one too were therefore a critical step towards this. Our integrative project was aimed at including the modelling and biomedical communities. We strived to be transparent about our internal processes and quickly publish our curation results. Reproducibility of scientific results was an inherent goal of this project, including replication of model code by other institutions, but also transparency of the workflow that we applied as well as publication of our “lessons learnt” as a guideline to measure reproducibility. In the future, BioModels will apply the updated curation pipeline whenever possible.

The University Medicine Greifswald, who was also responsible for contacting the different user communities, announcing the project on twitter, advertising the project in the biomedical communities and at relevant conferences.

The main project outcome is a collection of reproducible COVID-19 models, fully featured and provided as COMBINE archives. As side products of the project, we released new versions of the main software tools and libraries necessary to simulate SBML models and to generate COMBINE archives. The experiences from this project supported the publication of a reproducibility scorecard. The resulting models qualified to be integrated into the COVIDGraph, an open source knowledge graph on COVID-19 information.