Modeling support, curation, and dissemination to enable reproducibility
The Center for Reproducible Biomedical Modeling is an NIBIB/NIGMS/NSF funded center to encourage the publication of reproducible models. Studies indicate that the bulk of published peer-reviewed computational models in physiology and systems biology cannot be reproduced from the description in the published pape. Even if the code used in the study is made available it can be dififcult to reproduce the reported findngs. The center aims to encourage best practices, by working with journals, and developing technology to assist authors in publishing reproducible models. The lack of reproducibility can slow down research progress by preventing the dissemination of research findings to the broader biomedical community.
In addition, better reproducibility will enable larger and more accurate systems biology models, as well as their application to science, bioengineering, and medicine, by enhancing their understandability, and reusability. The center also aims to develop new standards and technology, community services, and provide training to the next generation of modelers.
Reproducible models provide important benefits, as they are can be readily understood, trusted, modified, and reused for additional analyses or to compose new models. Building models which are reproducible requires that:
- The modeler records the data, code, and decisions used to make and simulate models
- Comprehensible languages, using standard data formats and nomenclature, are used to describe the model
- Tthe artifacts produced during the modeling workflow be publicly shared using an open-source license
To support reproducible modeling, the Center provides a curation service led by Curation Service Director, David Nickerson, at the Auckland Bioengineering Institute. Modelers can submit a request for curation assistance for their model. A curator will create a private workspace for the submitted model, annotate the meaning and provenance of the model, reproduce reported figures through simulation of the model, and compare model predictions with reported outputs. A checklist and report will be returned to the modeler, evaluating the completeness of submitted resources and whether the model could be reproduced. The curator will submit the curated model, annotations and curation reports to a model repository following the peer-review process and publication. PLoS Computational biology recently partnered with the Center for Reproducible Biomedical Modeling to provide this service to modelers who submit to their journal.