Top Level Guide

The Brain Scaffold Builder is a framework that allows you to create workflows to reconstruct and simulate neural network models.

A typical workflow of the BSB works as follows:

1. The user provides a configuration that describes the network they want to reconstruct and simulate.

   It is interpreted by the BSB as a series of tasks to perform.

2. The BSB creates the topology of the network (i.e., its shape and size).

3. The BSB places cells within the network, following a defined strategy.

4. The BSB connects the aforementioned cells according to connectivity rules.

5. The final circuit is simulated with an external simulator, with the BSB serving as an intermediate to deal with the model.

For each step of this workflow, the BSB also provides a list of predefined strategies and tools to create your own with as little difficulty as possible. Hence, for most of your reconstructions, you may not have to code!

The BSB acts here as a black box with an extended interface to allow you to properly parametrize it.

Scaffold

The Brain Scaffold Builder revolves around the Scaffold object. A scaffold ties together all the information in the Configuration with the Storage. The configuration contains your model description, while the storage contains your model data, like concrete cell positions or connections.

Using the scaffold object, one can turn the abstract model configuration into a concrete storage object full of neuroscience.

To do so, the configuration leverages configurable objects to describe the underlying neural network, called components. Components define which methods and parameters should be used to reconstruct and simulate the network. The ones that you will probably employ the most are:

• Topology defines the shape and volume of your network, (it is composed of Regions and Partitions),

• Cell Types allows you to estimate the cellular composition (and attach Morphologies when needed),

• Placement places cells,

• Connectivity connect cells,

• Simulation simulates the resulting network. Each simulation consists of:

  • Cell Models describe how to simulate a cell.

  • Connection Models describe how to simulate cellular connections such as synapses and gap junctions.

  • Devices describe the experimental setup by ways of input stimuli, recording devices, LFP probes, etc.

Assembled together these components form a linear workflow that will build your network from scratch.

Through this interface, lies the ultimate goal of the entire framework:

To let you explicitly define every component that is a part of your model, and all its related parameters, in such a way that a single CLI command, bsb compile, can turn your configuration into a reconstructed biophysically detailed large scale neural network.

Configuration

The Configuration object is organized as a hierarchical tree. From the root, the main blocks branch off, consisting of nine main components:

• network

• storage

• regions

• partitions

• morphologies

• cell_types

• placement

• connectivity

• simulation

These blocks contain nested sub-blocks that form the network. Additionally, there are two optional components: after_placement and after_connectivity, where users can define specific hooks to run within the workflow. All these components will be described in more detail in the following sections.

The configuration object contains only the description of the model, not its implementation (python code) nor its data (stored in the storage object). It can therefore be stored in a separate file (usually Json or Yaml) that can be easily interpreted by the BSB.

What is next?

We are now going to introduce the different components through a tutorial, explaining how to build your first network .