FSKX-Cloud-Platform: A Microservice-Based Solution for Scientific Model Execution

The FAIR Scientific Knowledge Exchange (FSKX) format has established itself as a valuable standard for the exchange of data and models across scientific domains. Originally developed as the Food Safety Knowledge Exchange format, it has evolved beyond the food safety domain to facilitate adoption across multiple scientific disciplines. The Risk Assessment Knowledge Integration Platform (RAKIP[1]) Initiative has developed various software tools to drive FSKX adoption by model creators worldwide. Over the past years, we have encouraged model creators to convert their work into the FSKX format and publish them in our FSKX model repositories. The RAKIP-Web model repository saw a particular increase in attention, receiving highly curated models from various partners within the EU. We tried to make use of the interoperability that FSKX provides, by connecting several online journals and repositories to RAKIP-Web. For example, publishing an FSKX file in Zenodo[2] through our RAKIP-Web publication tool automatically generates a URL inside the Zenodo entry that links back to RAKIP-Web – directly allowing the online execution of the model with user defined simulation parameters (Figure 1). Providing a way to execute FSKX models online is one of the greatest features we can offer to model creators and scientists worldwide. However, as the number of models in our repositories increases, the limitations of our infrastructure have become apparent. Recent models sit side-by-side with models created years ago, with software dependencies that conflict with each other. To guarantee a functional runtime environment for each model, we needed to find a more robust solution. Preferably a solution that also enables 3rd party developers to access our FSKX services, because we also realized that our user interface can’t accommodate all our partners use-cases. To address these challenges, we present the FSKX-Cloud-Platform—a cloud-based solution that enables the execution of any FSKX model in tailor-made container environments. For each FSKX model, a container image is created with the exact software dependencies and runtime environment required. Through a REST API and event-driven architecture based on the EPCIS 2.0 standard, the appropriate container is deployed on demand, and model simulations are executed securely and efficiently. This white paper details the architecture, implementation, and benefits of the FSKX-Cloud-Platform, demonstrating how it enhances scientific reproducibility, improves user experience, and facilitates broader adoption of the FSKX standard. [1] https://foodrisklabs.bfr.bund.de/rakip-initiative/ [2] https://www.zenodo.org