Neurodamus

Neurodamus is a BBP Simulation Control application for Neuron.

The Python implementation offers a comprehensive Python API for fine tuning of the simulation, initially defined by a BlueConfig file.

Description

Neurodamus is the BBP in-house developed application for setting up large-scale neuronal simulations. It has traditionally been implemented as a set of extensions to Neuron, in the form of .hoc and .mod files. The parameters of the simulation are loaded from a configuration file, by default BlueConfig.

To address several limitations of the Hoc implementation, including development effort, the high-level layers of Neurodamus have been reimplemented in Python. Such implementation effectively makes available to the user a Python module with a comprehensive API, suitable to fine control simulation aspects, as well as inspect and eventually adapt the simulations as intended.

Install

Prerequisites

• hdf5
• libsonatareport https://github.com/BlueBrain/libsonatareport
• neuron with MPI https://github.com/neuronsimulator/nrn

Install neurodamus

git clone https://github.com/BlueBrain/neurodamus.git
cd neurodamus
pip install .

Build special with mod files

mkdir mods
cp -r mod-files-from-released-circuit mods/
cp -r neurodamus-install-prefix/share/mod/* mods/
nrnivmodl -incflags '-I <include-paths-of-our-dependencies>' -loadflags '-L <libs-paths-for-linking>' mod

Examples

Once installed, you should be able to find neurodamus in your path:

$ neurodamus
  Usage:
      neurodamus <ConfigFile> [options]
      neurodamus --help

Among the options you will find flags to tune run behavior.

Neurodamus explicitly depends on MPI libraries for parallel execution. Therefore please use "srun" or "mpiexec" to launch it, according to your platform. If you don't, complicated error messages may show up. Please remember it.

Even though a neurodamus launcher is provided, for production runs we suggest using special instead. This way has proven to take advantage of optimized math libraries. We hope to bring the same advantages to the launcher script soon.

srun <srun params> <your_built_special> -mpi -python $NEURODAMUS_PYTHON/init.py <neurodamus params>

An example of a full installation with a simulation run can be found in the workflow test simulation_test.yaml.

Docker container

Alternatively, you can start directly a neurodamus docker container where all the packages are built. With the container, you can build your mod files and run simulations. See instructions in docker/README.md.

Building the docker container

The docker container image is built automatically when a new tag is created in the repository, if a container with the specified version doesn't already exist. On manual pipeline runs, the container image is also built but not automatically pushed to docker-hub; you'll have to manually start the job in the pipeline you created if you want this to happen. Keep in mind that this is optional, the container image is always pushed to the gitlab registry!

The docker images will be built in the regular gitlab pipeline - if triggered under the right conditions (either manually or through git tag creation) the container jobs will be added to this pipeline.

If you run the pipeline manually, you can also set versions for the dependencies:

• LIBSONATAREPORT_TAG: git tag for libsonata-report
• LIBSONATA_TAG: git tag for libsonata
• NEURON_COMMIT_ID: commit ID for neuron
• REGISTRY_IMAGE_TAG: this will be the container version. Must be specified if you specify one of the others.

Acknowledgment

The development of this software was supported by funding to the Blue Brain Project, a research center of the École polytechnique fédérale de Lausanne (EPFL), from the Swiss government's ETH Board of the Swiss Federal Institutes of Technology.

Copyright (c) 2005-2023 Blue Brain Project/EPFL