What can I find here?

• Description of TVB GUI Widgets
• Installation and usage guidance
• Basic usage examples

Authors

• SDL Neuroscience Juelich: Sandra Diaz (lead)
• INS Marseille: Jan Fousek (PO)
• Codemart: Paula Prodan, David Bacter, Romina Baila, Teodora Misan, Rares Horge, Jochen Mersmann, Lia Domide (technical team)

Concepts

• GUI, Widgets
• JupyterLab Extensions
• data-oriented arch, API, DataTypes
• HPC backends and monitoring
• KG integration
• GUI workflow editing

MOTIVATION

The showcases developed in the last phase of the HBP are meant to illustrate the full potential of technical and scientific features offered by EBRAINS. In order to support the usability of the showcases, as well as future EBRAINS workflows, we have developed a set of modular graphic components and software solutions, which can be easily deployed or replicated in the EBRAINS Collaboratory within the JupyterLab.

These graphical user interfaces (GUI) components are all based on and under open source licences, supporting open neuroscience and they enable features like:

• Easy setup of models and region specific or cohort simulations
• Selection of Data sources and their links to models.
• Querying data from Siibra and the Knowledge Graph.
• Deployment and monitoring jobs on HPC resources.
• Integration of a subset of the virtual brain analysis and visualisation tools.

METHODS

The graphical components developed, called tvb-widgets, are designed to be integrated into cells of notebooks or are part of JupyterLab extensions for direct usage in the EBRAINS Collaboratory as independent panels  with an appealing look and feel (3D display and interaction, reactive UI, drag&drop events, etc).

The solution is modular, easy to extend and applicable to multiple showcases within EBRAINS.

For the development of these modules, we choose a Data Centred Architecture, where data is annotated, and accessed independently by the satellite components which can read or modify it. We introduce a hierarchical representation with flexible types for data and its metadata; the components access shared data structures and are relatively independent —they interact only through data exchange; we get automatic orchestration as a possibility.