Simulation

Solutions for brain researchers to conduct sustainable simulation studies and share their results

Forschungszentrum Jülich / SBC Lehmann

Aix-Marseille Université / Viktor Jirsa

Morphologies:Markram, H.,et al. (2016). Zenodo. doi: 10.5281/zenodo.57082; Mercer A, Thomson AM. Front. Neuroanat. doi:10.3389/fnana.2017.00083; Morphologies and model: Migliore et al. (2018). PLoS Comput Biol; doi:10.1371/journal.pcbi.1006423

In research, simulation acts as the main conduit for interchange between experiment and theory. It is a powerful instrument in our endeavour to understand the human brain, which is a complex dynamic system with a multi-scale architecture, further complicated by significant differences between one person’s brain and another’s. The complexity and versatility of the brain, and the variations from one brain to another, are major scientific challenges, driving the development of simulation technology

Neuroscientists have different views on how to best tackle the intricacy of complex systems.

Neuroscientists have different views on how to best tackle the intricacy of such complex systems, advocating approaches that range from holistic to minimalistic, from the notion that only realistic models can account for the inner workings of our brains, to the assumption that only systematic model simplification will allow us to uncover fundamental principles. Likewise, brain models differ in size, complexity and level of detail. EBRAINS Simulation services offer technical solutions for brain researchers to conduct sustainable simulation studies and build upon prior work, and the means to share their results. The services provide integrated workflows for model creation, simulation and validation, including data analysis and visualisation. The simulation engines cover the entire spectrum of levels of description ranging from cellular to network to whole brain level.

Simulation

Services

Cellular level simulation

The EBRAINS Cellular Level simulation tool suite includes an ensemble of custom workflows allowing neuroscientists to design, build, simulate and visualise neuronal models; from individual cells to entire brain regions, with cellular-level details. The cell and tissue model construction covers different brain areas and is driven by experimental data that allows in silico exploration of their computational properties and their role in health and disease.

    • Simulation

    NeuroScheme

     Create, navigate, explore and interact with schematic representations of neurons and circuits
    • Simulation

    NeuroTessMesh

     Visualise neural anatomy in great detail
    • Simulation

    ViSimpl

     Study simulation data and their associated spatial and temporal features by exploratory visual analysis
    • Simulation

    Arbor

     High-performance library for computational neuroscience simulations, ranging from single-cell models to large networks
    • Simulation

    Interactive Workflows for Cellular Level Modeling

     Build, reconstruct, and simulate data-driven brain models
    • Simulation

    NEURON /CoreNEURON

     A simulator for modelling individual neurons and networks of neurons
    • Simulation

    NetPyNE

     Build, simulate, and analyze data-driven multiscale brain circuit models
    • Simulation

    Snudda

     Leverage your neuron morphologies and experimental data to create microcircuits with realistic connectivity

Network level simulation

Brain function emerges at the network level. Even basic functions such as the initial processing of visual input require the interaction of many neurons in different brain areas. Network-level simulation technology enables investigation of such brain-scale networks, covering multiple brain areas at the resolution of single neurons and synapses. The direct link to brain function makes network-level simulation attractive for AI and robotics applications. To enable brain-scale simulations, the technology needs to be extremely scalable and capable of exploiting modern supercomputers. The simplifications that underpin network-level simulation technology have also inspired the design of neuromorphic systems.

    • Simulation, Brain-Inspired Technologies

    Neuromorphic Computing

     Simulate or emulate spiking neural networks with neuromorphic compute systems
    • Simulation, Brain-Inspired Technologies

    Neurorobotics Platform

     Simulating intelligent agents within realistic environments
    • Simulation

    Elephant

     Analyse the neuronal dynamics of experiments and brain simulations with this Electrophysiology Analysis Toolkit
    • Simulation

    NEST Desktop

     Construct neuronal networks and explore network dynamics with the NEST Simulator GUI
    • Simulation

    NEST Simulator

     A simulator for spiking neural network models of any size
    • Simulation

    NESTML

     A domain-specific language to write custom neuron models
    • Simulation, Brain-Inspired Technologies

    PyNN

     A simulator-independent language for building neuronal network models
    • Simulation

    Brain Scaffold Builder

     Flexible and organized workflows for neural circuit reconstruction and simulation at different levels of detail

Whole-brain level simulation

To simulate whole brains, the white-matter axon fibre bundle network, the so-called structural connectome, is reconstructed to define the model network. To reduce complexity and computational demands, the activity of brain regions is typically not simulated by networks of individual neurons. Instead, mean-field theory is used to replicate the main dynamics of large groups of neurons. The resulting models of the different networks are connected by the weights and time delays of reconstructed structural connectomes. Whole brain simulations have been used to explain several empirically observed phenomena, such as the emergence in fMRI of functional networks, the spread of epileptic seizures or relationships between phenomena observed in fMRI and electrical neural activity over several temporal scales. In EBRAINS, the simulation tools at the whole brain level are developed around the framework of The Virtual Brain (TVB).

    • Simulation

    The Virtual Brain

     Create personalised brain models and simulate multi-scale networks

Data analysis and visualisation

Data analysis provides methods and workflows to reveal hidden dynamics and extract characteristic statistical features obtained from simulations and experiments. These toolboxes help computational neuroscientists analyse simulations of neuronal dynamics and assist in validating models against experimental evidence. Visualisation provides scientists with a powerful tool for interactive data analysis, from an overview to deep insights. EBRAINS offers access to interactive tools for visualisation of models and simulation results, from the cellular to the network level.

    • Simulation

    NeuroScheme

     Create, navigate, explore and interact with schematic representations of neurons and circuits
    • Simulation

    NeuroTessMesh

     Visualise neural anatomy in great detail
    • Simulation

    ViSimpl

     Study simulation data and their associated spatial and temporal features by exploratory visual analysis
    • Simulation

    Elephant

     Analyse the neuronal dynamics of experiments and brain simulations with this Electrophysiology Analysis Toolkit
    • Simulation

    NEST Desktop

     Construct neuronal networks and explore network dynamics with the NEST Simulator GUI

Molecular and subcellular level simulation

Roadmap

2021, Q2

PyNN represented on EBRAINS

Releases of NEST 3.0 and accompanying Education Frontend (NEST Desktop 3.0), and of NESTML 4.0 with ODE-toolbox 2.3

2021, Q3

Molecular and subcellular level simulation tools and services represented on EBRAINS

First use-case workflow for co-simulation using the standardized framework on local machines

Release of the first version of the visualization framework’s documentation including a description of an example curricula for the Education Frontend (NEST Desktop)

Release of PyNN 0.10.0 supporting NEST 3.0 and NEURON 8.0

Complete migration of BSP use cases to the cellular level simulation workflows and use cases page on EBRAINS

2021, Q4

Co-simulation services represented on EBRAINS, and HPC-deployed use case for co-simulation using standard framework available

Release of the second version of the visualization framework’s documentation

Alpha version of NRP and co-simulation workflow available

Improved cellular level simulation workflows and use cases to take into account user requests

Finalized workflow ensuring that all parts of NEST documentation are complete

2022, Q2

Support for third-factor learning rules in NESTML

Releases of PyNN 1.0.0 (stable API) and Elephant 1.0 (stable API and visualization)

Upgraded cellular level simulation workflows and use cases for interaction with EBRAINS

Initial integration of TVB with the data representation and communications standard for simulation defined in EBRAINS for co-simulation on HPC

2022, Q3

Release of the visualization framework with documentation including tutorials and integration into EBRAINS

Common user-level documentation for data analysis and model-validation components available

2022, Q4

Simulation and analysis workflows for each simulation scale and co-simulation, including integration in EBRAINS infrastructure, documentation, validation, visualization where appropriate, integration testing and user support workflows

Completed use-case based validation of visualization framework

2023, Q1

NESTML supporting SpiNNaker platform code generation

Release of PyNN 2.0.0 supporting multi-compartmental modelling with NEURON and Arbor

Two-way translation and control module ready for neural mass to point-neuron models

Access all EBRAINS services