A team led by Lucian P. Smith of the Center for Reproducible Biomedical Modeling, together with Center Director Herbert M. Sauro and collaborators from institutions around the world, has published “Verification and reproducible curation of the BioModels repository” in PLOS Computational Biology. Supported by the Center, the study addresses a major challenge in computational biology: ensuring that published model results can be reliably reproduced and verified by other researchers.
The BioModels repository contains more than 1,000 curated computational models of biological systems. Although these models are generally stored using the standardized Systems Biology Markup Language, or SBML, that format describes the model itself but does not necessarily include the detailed instructions needed to recreate a specific simulation or published result. The Simulation Experiment Description Markup Language, or SED-ML, provides a standardized, software-independent way to record those instructions.
The research team updated, corrected, or created SED-ML files for 1,055 curated models and tested them across five different biosimulation engines. Almost all of the previously existing SED-ML files required at least a minor correction. Following this extensive curation and testing process, 932 models, or 88 percent, produced consistent results using at least two simulation engines.
This work substantially improves the reproducibility, credibility, and usefulness of the BioModels collection. The project also identified and corrected problems in models, simulation instructions, software wrappers, and simulation engines, strengthening the broader infrastructure used for computational biomedical research and making these models better suited for reuse in future studies.
Read the full article:
https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1013239