Verification and reproducible curation of the BioModels repository

Area of Science:

• Computational biology
• Systems biology
• Bioinformatics

Background:

• The BioModels Repository houses over 1000 curated mechanistic models, primarily in Systems Biology Markup Language (SBML).
• SBML formally defines models but lacks computational experimental conditions, hindering result reproducibility.
• Simulation Experiment Description Markup Language (SED-ML) standardizes experimental conditions for simulations.

Purpose of the Study:

• To enhance the reproducibility and credibility of curated mechanistic models in BioModels.
• To update and correct existing SED-ML files and provide new ones for a substantial portion of the BioModels collection.
• To leverage SED-ML for verifying simulation results across multiple computational engines.

Main Methods:

• Updated and corrected SED-ML files for 1055 curated mechanistic models from the BioModels Repository.
• Developed a wrapper architecture for interpreting SED-ML files.
• Performed verification of simulation results across five different Ordinary Differential Equation (ODE)-based biosimulation engines.

Main Results:

• Successfully updated and corrected SED-ML files for 1055 BioModels entries, addressing previous inaccuracies.
• Demonstrated the implementation-independent nature of SED-ML through cross-engine verification.
• Provided a more robust and reliable set of simulation experiment descriptions for a large collection of biological models.

Conclusions:

• The updated SED-ML files significantly improve the reproducibility of results from BioModels.
• Cross-platform verification using SED-ML enhances the credibility of computational models.
• This work strengthens the utility of the BioModels collection for the systems biology community.