Work Package 5

EBRAINS Modelling Services

What we do

This Work Package manages the overall architecture integrating scientific software into EBRAINS. It comprises the development and advancement of applications for brain modelling and simulation, analysis of neural activity data, model validation, and visualisation. The deployed services will combine the applications into comprehensive workflows, they will enable multiscale co-simulation of brain processes and closed-loop neuroscience experiments/in silico/, which in turn support R&D on embodied AI and robotics. WP5 delivers the e-infrastructure for operating all services of the Project on the computing infrastructure allocated to them.

How we are organised

WP5 comprises Service Category 3 (SC3) - Brain modelling and simulation workflows: integrated tools to create and investigate models of the brain, SC4 - Closed loop AI and robotics workflows: design, test and implement robotic and AI solutions tasks, and tasks that will deliver the necessary e-infrastructure for all service categories of the Project.

Tasks

Task T5.1 Molecular and subcellular level description (SC3)
Task T5.2 Detailed cellular level description (SC3)
Task T5.3 Spiking network level description (SC3)
Task T5.4 Whole brain level description (SC3)
Task T5.5 Multi-simulator interaction (SC3)
Task T5.6 Analysis and validation framework (SC3)
Task T5.7 Visualisation framework (SC3)
Task T5.8 Neurorobotics back end (SC4)
Task T5.9 Neurorobotics front end (SC4)
Task T5.10 Neurorobotics closed-loop models (SC4)
Task T5.11 EBRAINS Technical Coordination
Task T5.12 EBRAINS Infrastructure Middleware
Task T5.13 EBRAINS Implementation Management
Task T5.14 EBRAINS Integration, Provisioning and Operations
Task T5.15 WP5 Coordination and Management, including supervision of SC3 and SC4
Task T5.16 The Functional Interplay between Plasticity Processes for Arbor (FIPPA) (SC3)
Task T5.17 Molecular level tools for multiscale simulations (SC3)
Task T5.18 Arbor Implementation of the Inferior Olive Network (ArborIO) (SC3)
Task T5.19 Allosteric modulation of pentameric ligand-gated io channels: relevance to covid 19
Task T5.20 Personalized human musculoskeletal model and exoskeleton coupling with simulated neurorehabilitation
Task T5.21 Closed-loop neurorehabilitation simulator for upper limb
Task T5.22 Blocking SARS-CoV-2-mediated neurodegeneration and brain damage via NLRP3 inflammasome inhibition

Publication highlights

Gaute T. Einevoll, Alain Destexhe, Markus Diesmann, Sonja Grün, Viktor Jirsa, Marc de Kamps, Michele Migliore, Torbjørn V. Ness, Hans E. Plesser, Felix Schürmann, The Scientific Case for Brain Simulations, Neuron, Volume 102, Issue 4, 2019, Pages 735-744, ISSN 0896-6273, https://doi.org/10.1016/j.neuron.2019.03.027.

Galindo Sergio E., Toharia Pablo, Robles Oscar D., Pastor Luis, ViSimpl: Multi-View Visual Analysis of Brain Simulation Data, Frontiers in Neuroinformatics, 10, 2016, 10.3389/fninf.2016.00044

Contact

Work Package Leader: Yannis Ioannidis

Work Package Manager: Evita Mailli

Webpage editor: wp5(at)humanbrainproject.eu

A workflow for constructing and optimizing data-driven models of subcellular signalling

We have created a software tool-set that helps with creation of models, storage of models, auto-generation of model code, parameter estimation for those models (using lab data), global sensitivity analysis of the model with respect to the parameters, and model based prediction.

https://github.com/icpm-kth

A set of fully operational workflows for cellular level in silico experiments on single cell, small microcircuits, and large-scale circuit models, which will be fully integrated into EBRAINS, complemented by tutorials/user guides, and publicly available for users

website is online and all workflows are operational

https://www.ebrains.eu/tools/interactive-workflows-for-cellular-level-modeling

Open-source software releases of NEST featuring third-factor learning rules, long axonal delays, revised code for sustainability, co-simulation and visualization interfaces, simulation performance, performance in creation of hierarchical networks, documentation and training material

NEST simulator is continuously developed and was released regularly during SGA3 (see https://github.com/nest/nest-simulator/releases). Contribution of T4.13 (via vouchers).

Open-source software releases of PyNN biophysical neuron API (distributed simulations of network models with biophysically/morphologically-detailed neurons, running on both NEURON and Arbor, with support for import/export in SONATA format and comprehensive documentation)

PyNN 0.12.0 contains a beta version of the biophysical neuron API, with support for NEURON and Arbor. Some example scripts using the new API are available at https://github.com/apdavison/PyNN-mc-examples

Open-source software release of Model Validation Test Suites supplying a framework for validation tests to perform model validation at various levels including documentation

Six validation libraries are available (HippoUnit, MorphoUnit, NetworkUnit, BasalUnit, CerebTests, eFELUnit) to support structured model validation by providing (i) validation test definitions and (ii) interface definitions intended to decouple model validation from the details of model implementation. All of the libraries are based on the same SciUnit framework, making them easily to integrate into the Model Validation Web Service (OP5.16). The libraries have received incremental updates and releases during SGA3 that increase and improve functionality and add further unit testing.

Open-source software releases of Elephant including data analysis object models and refined documentation

Recent Elephant releases have improved the software through new functionality (e.g., packages for performing multitaper and Granger causality estimation) and new features in existing methodology, bug fixes and continuous updates in the content and consistency of the documentation. Elephant tutorials can now be launched directly in the EBRAINS Lab from the documentation. In addition, a data analysis model for describing analysis provenance was developed and integrated into a solution for online provenance capture. For technical reasons, this model was externalized as a separate software package Alpaca (see https://alpaca-prov.readthedocs.io for concrete examples on capturing data analysis outputs of Elephant). In addition, a new model for handling experimental trials was developed and will be contained in the upcoming Elephant release.

Open-source software releases of Elephant Visualization delivering visualizations for most Elephant functions

The Viziphant library provides a range of functions to visualize electrophysiological data and analysis outputs obtained using the Elephant tool, focusing on complex analyses where outputs require specialized visualizations, such as higher-order pattern search analyses or visualizations of neuronal trajectories. Release notes are available at https://viziphant.readthedocs.io/en/latest/release_notes.html.

https://viziphant.readthedocs.io/en/latest/release_notes.html

Open-source software releases of Model Validation Web Service including improved browsing and presentation of validation results

We have made regular releases of improvements to the Model Validation web services, including: - rewritten graphical app with EBRAINS authentication and Collaboratory v2 integration - improved homepage with highlighted models - curation dashboard interface for model curators - addition of end-to-end testing using Cypress - bug fixes and performance improvements - migration from Knowledge Graph (KG) v2 to KG v3 with openMINDS schemas.

Open-source software releases of the Education Front end (EduFro) - integration with the SimDaVis, connection to SimVisSuite, curricula, user interface evaluation, high software maturity, documentation and training material (for teachers)

The list of aspects mentioned in the title of the OP has been achieved.

Open-source software releases of Neo including documentation

We have made seven releases of new versions of Neo on the Python package index, with documentation at https://neo.readthedocs.io. Neo is also available as part of the EBRAINS software distribution.