Empowering Neuroscientists: New Open Source Tools for Modeling

© 2023 EPFL

© 2023 EPFL

In conjunction with the recent release of preprints on the modeling of the rat Somatosensory Cortex, the brain’s center for processing sensory information, EPFL’s Blue Brain Project has open sourced important software packages for simulation neuroscience. It is hoped that access to these new tools will help neuroscientists make the most of the large amounts of data as well as the related open source model which BBP researchers have shared at the same time.

Concurrently with the release of a two-part preprint on the juvenile rat non-barrel Somatosensory Cortex (nbS1) covering both its anatomy and physiology as well as the release of the model itself, Blue Brain is sharing three complementary software packages. These were central in the simulation and analysis of the nbS1, a model which contains over 4 million structurally detailed neurons and more than 14 billion connections. Working with a model of that scale and level of detail is extremely complicated; Blue Brain therefore developed software tools for the simulation, validation, and analysis of the model. The tools work on top of the open Sonata standard developed for the representation of complex neuronal networks, and as such are compatible with any Sonata model. These tools, developed and used internally for the research of the preprints, are being made available and can be used to calibrate any circuit, run simulations, and analyze results.

ConnectomeUtilities is a powerful tool for researchers working with complex networks; it offers streamlined analysis and provides insights into network properties, especially in the context of biologically detailed models of neural circuits. BlueCelluLab is a versatile tool that allows scientists to conduct virtual experiments, helping them better understand the intricate behavior of brain cells. Meanwhile, Neurodamus enables the running of large-scale simulations on powerful computer clusters efficiently.

ConnectomeUtilities: a software package to simplify topological analyses on complex networks

A connectome is the equivalent of an extremely large and complicated brain map, representing how all the brain cells are connected together. ConnectomeUtilities is a tool that helps scientists analyze and understand these brain maps better. It doesn't create the maps, but it makes it much easier to study them. On the one hand, it simplifies the maps so that important details are kept, but they become easier to work with. On the other, it is also able to automate some common tasks, like comparing different parts of the brain or creating control models to compare against. The tool is not limited to just one type of brain map; it can work with different types of brain data. While a researcher has to provide the analysis they want to run, ConnectomeUtilities makes it easier to apply it to detailed brain models. "Researchers already know very well how to analyze complex networks when they are represented as a graph, as nodes with single connections between them. But the brain is more complicated, with multiple, constantly changing connections of different types and strengths between neurons. ConnectomeUtilities helps us deal with that in systematic and reproducible ways" points out Blue Brain scientist Michael Reimann. In addition to its main "ConnectivityMatrix" class, the software provides various other functionalities, such as comparing results to random controls to understand their significance, representing connectivity at lower resolutions, and efficiently calculating path distances between synapses on neuronal morphology in a model. It can also help with the analysis of plasticity, simulations of how connections change over time or how synaptic connections are affected by the physical structure of the brain. Overall, this package helps scientists make sense of complex brain maps, so they can better understand how the brain works.

BlueCelluLab: a powerful tool to conduct virtual experiments on small groups of cells in the brain

When modeling a very large circuit of brain cells, it can be relevant to study only a small group of cells within it. “To understand how the entire brain works, we first have to understand how every small part of it works. As such, even when we have a large model available, we sometimes pick and simulate only a small group of cells within it” clarifies Michael Reimann. BlueCelluLab acts like a magnifier, taking the output of large simulations and focusing on that specific group of cells while keeping their connections to the rest of the circuit intact. This way, the chosen cells can be modified and experimented on while remaining appropriately embedded within the model, without having to rerun the entire large simulation again. For example, you can inject current into one part of a cell and see how it affects other parts, or connect two cells to observe how one cell's activity impacts the other. Researchers can even run these manipulations and smaller simulations on their laptops, thereby making the most out of large simulations of the entire model. The package allows the creation of detailed reports on specific aspects of cells after the simulation, as well as validating the properties and dynamics of the connections over time and developing rules for how these connections change and can automate virtual experiments mimicking whole-cell patch clamp experiments. “The BlueCelluLab package is particularly useful when troubleshooting scientific and computational aspects of a simulation”, says Scientific Software Developer Anıl Tuncel who worked on the software. BlueCelluLab was used to calibrate and validate synaptic physiology parameters of the nbS1 circuit.

Neurodamus: a control program reading circuits and instantiating simulations

Neurodamus is a simulation control application developed by Blue Brain for setting up large-scale simulations of neurons. It’s built on top of the widely used NEURON simulator and can use CoreNEURON for efficient execution. “The high-level components of Neurodamus are written in Python and with a user-friendly interface that can be easily extended” adds High Performance Computing Engineer Pramod Kumbhar, “Using NEURON API, it abstracts all parallelization details and provides an easy way for researchers to configure and run simulations.” Neurodamus was used to run all network simulations behind the nbS1 study.

In summary, all three software packages can be used to analyze, validate and simulate any Sonata circuit model. Connectome Utilities proves to be a valuable tool for researchers dealing with complex networks. It simplifies analyses and provides insights into network properties, especially in the context of detailed neural circuit models. The BlueCelluLab package is versatile and empowers scientists to conduct virtual experiments, aiding their understanding of brain cell behavior. Finally, Neurodamus facilitates large-scale simulations on powerful computer clusters, ensuring efficient processing.

“We hope that these tools will help other researchers to make the most out of the over 50 GB of model and data that Blue Brain has shared via the Zenodo platform,” emphasizes Blue Brain Scientist András Ecker. “We believe the introduction of these software marks a new phase for open Blue Brain datasets, offering exciting possibilities for the neuroscience community” concludes Blue Brain founder and Director, Prof. Henry Markram.