MMODA multi-messenger online data analysis platform in the frame of the EuroScienceGateway project

The platform makes use of cloud-based data management solutions and virtualization technologies. It is intended to meet the challenges of efficient sharing and reuse of data, long-term preservation of data and data analysis workflows and the reproducibility of results. Some analysis modules (like the one for the INTEGRAL data analysis) developed as separate plugins. In addition to these “fixed” plugins, the analysis workflows in the form of Jupyter notebooks, annotated by the terms from our platform-specific ontology, can be automatically deployed as a service with web-GUI and REST API. 

The MMODA platform is based on a federated computing environment. Each instance is based on a layer of physical or virtual machines running microservices in the containerized form. For the deployment, we use Kubernetes container orchestrator, which is automatically deployed by the Rancher Kubernetes Engine (RKE2) solution via the TerraForm module on top of OpenStack IaaS cloud of the French academic cloud federation, FG-cloud. Helm charts are provided to deploy all MMODA services (web interfaces and API, task manager, adapters corresponding to scientific analyses, etc.). We use the GitOps practices to deploy the platform software stack.

Many approaches to data management and workflow publishing are rather close between communities, but realisations are completely separate. As an example, the BioInformatics community has somewhat widely adapted Galaxy platform, enabling open data analysis services with distributed compute. We joined (along with particle physics, material science, climate science) an EuroScienceGateway EOSC project in order to bring our tools and workflows into Galaxy platform, developing synergy between MMODA and Galaxy. We explore peculiarities of the astroparticle-oriented workflow publishing on the WorkflowHub platform. This project, as follows, allows minimizing redundancy and enable cross-domain knowledge transfer.