More comprehensive and more predictive models of cells could enable the advancement of biology, medicine, and bioengineering. Such models will require collaboration between many investigators and domains. In turn, models must be shareable, executable, and accessible. While many community resources exist, such as BiGG, BNGL, BioModels, CellML, KiSAO, NeuroML, OMEX, SBML, and SED-ML, significant barriers to collaboration remain. Different types of models remain siloed across multiple repositories, modeling frameworks, model formats, simulation algorithms, and simulation tools. For example, flux balance models are often obtained from BiGG and simulated with COBRApy, while deterministic kinetic models are often obtained from BioModels and simulated with tellurium. Future, more comprehensive models will likely require capturing biology at multiple scales, requiring the ability to combine frameworks, formats, algorithms, and tools. Additionally, installation and usage of tools cam be difficult. To address these challenges, we have developed BioSimulations (https://run.biosimulations.org), a cloud platform for reproducing and reusing models with ease. BioSimulations will provide users a unified platform to share, discover, run, and visualize biological models across various modeling scales, formats, frameworks, and algorithms.
Recent advances Recent improvements to the platform have significantly increased the range of supported simulations to 10 model languages, 20 simulation tools, 70 algorithms, and several modeling frameworks including continuous and discrete kinetic, logical, flux balance, spatial, particle-based, and hybrid modeling. To help investigators navigate these tools, the platform can now automatically inspect models and recommend specific algorithms and tools for executing them. Large-scale simulation results up to 5 TB per project can now be saved and queried using the Highly Scalable Data Service (HSDS). Users can also use Vega to visualize simulation results with advanced, interactive diagrams, such as activity flow diagrams, process description maps, and reaction flux maps which can be created with a variety of tools such as Escher, GINSim, and Newt. Furthermore, BioSimulations makes it easy for investigators to reuse such visualizations across multiple projects, such as to compare multiple models or simulations.
Future directions We are developing a repository of entire simulation projects, including interactive visualizations of results.