Simulation Inputs

This page provides an overview of the nomenclature used for inputs in a Neurodamus circuit simulation. It assumes the reader is already familiar with the NEURON software and its terminology. Otherwise, here are the NEURON docs. For much more detailed information about the input files and the SONATA format, the reader can refer to the SONATA docs.

simulation_config.json

This is the main SONATA file that controls the simulation settings. Key fields include:

  • What reports the simulator should generate.

  • The duration of the simulation.

  • The Neuron stimuli.

A report is one of the possible outputs of the simulation. It typically consists of a time trace of a specific value (e.g., voltage) recorded per neuron. Reports can also be generated per neuronal section.

circuit_config.json

Defines how neurons are structured and connected. It can be seen as a graph of graphs, representing interactions between different neuron populations. This file references:

  • nodes.h5: Contains neuron properties. Note that these nodes are unrelated to neurodamus.node.Node, which is a low-level simulation orchestrator.

  • edges.h5: Defines connections between neurons.

A population is a group of neurons that interact within themselves and with other populations. NEURON does not have a concept of populations—this is specific to Neurodamus.

Neurons in nodes.h5 are primarily defined by their Electrical Type (EType) and Morphology Type (MType).

  • ETypes: Ion channel distributions that have been reverse-engineered from specific firing patterns. Assigning an ion channel distribution (distributed mechanisms) to a population often involves genetic algorithms and trial-and-error. They are provided as hoc files, and the path to the folder containing all EType hoc files for a circuit is specified in circuit_config.json. Individual neurons reference their assigned EType in nodes.h5.

  • MTypes: Categories of neuron morphologies (graph structures). Some neurons have extensive branching, while others feature a single prominent axon. NEURON itself does not recognize MTypes, as it requires instantiated neurons in memory rather than generic morphology categories. However, Neurodamus uses MTypes to facilitate circuit development.

  • Morphologies: Specific instances of MTypes. During circuit building, multiple neuron structures (graphs of sections) are generated from MTypes and stored in .asc (NeuronLucida) or .swc files. These morphology files are referenced in nodes.h5. circuit_config.json also specifies the directory containing these morphology files. Due to the complexity of generating many neurons, some morphologies may be reused.

node_sets.json

This file specifies which neurons (by GID) are included in a particular simulation. It overrides the default selection provided in circuit_config.json, which typically includes all neurons unless node_sets.json is supplied.

Source

inputs.rst