Building a first circuit

In this section, you will learn how to reproduce the different cerebellar cortex circuits built at the DBBS laboratory of the University of Pavia.

Introduction to BSB

The Brain Scaffold Builder (BSB) is a framework that provides tools and pipelines to reconstruct and simulate neural networks in silico.

BSB consumes a model description or configuration that describes the process of building a neural circuit from scratch. This includes the creation of its different sub-regions, the placement of the different cells that compose it and the connections of these cells to form an network. BSB implements also simulators backends that allow for the simulation of the produced circuits. Hence, the configuration can be extended to include simulation paradigms. Model configurations are stored as dictionaries, and usually loaded from yaml or json files.

A modular configuration file to reconstruct cerebellar cortex circuit

Configuration files will vary according to the specificities of each model, including the species, the sub-regions of interest, or the subject disease.

To help with the reproducibility of the different results from the DBBS we are here decomposing the configuration files used in our models in sub-configurations that can be assembled and/or switch on based on the model you wish to reproduce. This means that sub-configurations can be used to override parameters (e.g. modifying sub-region size to match disease conditions) or introduce new features to the circuit (e.g. adding a new cell type to the circuit). For each configuration file, we also associate a full description of its parameters, including their provenance and rationale behind them.

We have split our configuration based on the species studied. For each species, we also produced a default model or canonical circuit that will be used as a base for their alternatives. The other sub-configurations (that can be used to override or extend the default one) are called extensions.

Learn more about our configuration in the configurations section

Reconstruction of the rodent canonical circuit

As an example, we present here the process to reconstruct the cerebellar cortex of the rodent brain, based on the reconstruction of De Schepper et al. 2022 with BSB. This corresponds to our canonical circuit for the mouse so there is no need to assemble multiple configuration files.

Assuming you are in the cerebellar-models folder, run the following command in your terminal:

bsb compile configurations/mouse/mouse_cerebellar_cortex.yaml -v4 --clear

This command will produce a microcircuit of the mouse cerebellar cortex and store it in the mouse_cerebellum.hdf5 file. This process might take a while depending on your machine. This command will also automatically produce a separated pdf report on the structure of the microcircuit obtained: bsb_report_structure.pdf (see also the analysis section).

Congratulations you have built your first cerebellar cortex model !

Simulation of the reconstructed circuits

A list of simulations can be attached to a BSB reconstruction through its configuration file (read more here). These simulations are dependant on the simulator you choose and their paradigm. Some examples are provided for each simulator sustained by the cerebellar-models package in the configurations section.

After the simulation has completed, cerebellar-models allows you to run a list of predefined analysis on the simulation results. For instance, for spiking based simulation, you can run the following python script to produce a full pdf report with basic spike analysis of the simulation results (see more in spike reporting section):

from cerebellar_models.analysis.spiking_results import BasicSimulationReport

scaffold_file = "path/to/your_bsb_scaffold.hdf5"
simulation_name = "name_of_your_simulation"
nio_folder = "path/to/folder_containing_nio_files"
output_filename = "path/to/report_file_name.pdf"

report = BasicSimulationReport(scaffold_file, simulation_name, nio_folder)
report.print_report(output_filename)