Welcome to the Brain Modeling Toolkit

The Brain Modeling Toolkit (BMTK) is a python-based software package for building, simulating and analyzing large-scale neural network models. It supports the building and simulation of models of varying levels-of-resolution; from multi-compartment biophysically detailed networks, to point-neuron models, to filter-based models, and even population-level firing rate models.

_images/levels_of_resolution_noml.png

The BMTK is not itself a simulator and will utilize existing simulators, like NEURON and NEST, to run different types of models. What BMTK does provide:

  • A unified inteferce across different simulators, so that modelers can work with and study their own network models across different simulators without having to learn how to use each tool.

  • An easy way to setup and initialize network simulations with little-to-no programming necessacary

  • Automatic integration of parallelization when running on HPC.

  • Extra built-in features which the native simulators may not support out-of-the-box.

The BMTK was developed and is supported at the Allen Institute for Brain Science and released under a BSD 3-clause license. We encourage others to use the BMTK for their own research, and suggestions and contributions to the BMTK are welcome.

The latest release, previous releases, and current development can be found at https://github.com/AllenInstitute/bmtk

The BMTK Workflow and architecture

BMTK can readily scale to run models of single neurons, and even single compartments, and for all different types of neuronal networks. However BMTK was designed for very-large, highly optimized mammalian cortical network models. Generating the connectivity matrix could take hours or even days to complete. Plus once we had the base-line model we wanted to run against a large variety of conditions and stimuli, so that we could directly compare with existing in-vivo recordings (see Allen Brain Observatory).

_images/bmtk_workflow.jpg

Unlike other simulators, BMTK separates the process of building, simulating, and analyzing the results. First a fully instantiated base-line version of the model is built and saved to a file so each time a simulation is ran it takes only a small fraction of the time to instantiate the simulation. Results are also automatically saved to a disk. BMTK and the format it uses (SONATA, see below) makes it easy to dynamically adjust cell and synaptic parameters so that multiple iterations of a simulation be done as fast as possible

As such BMTK can be broken into three major components:

  • The Network Builder [bmtk.builder] - Used to build network models

  • The Simulation Engines [bmtk.simulator] - Interfaces for simulating the network

  • Analysis and Visualization Tools [bmtk.analyzer] - Python functions for analyzing and visualizing the

    network and simulation results

_images/bmtk_architecture.jpg

The components can be as one workflow or separately by themselves. BMTK utilizes the SONATA Data format (see next section) to allow sharing of large-scale network data. As such, it is possible to use the Network Builder to build the model but another tool to run the model. Or if another model has been built by someone else and saved in SONATA format BMTK will be able to simulate it.

SONATA

SONATA is a multi-institutional developed standardized, cross-platform data format for storing large scale networks and simulation results. The BMTK utilizes SONATA when building and simulating networks, so much of what is being described in the documentation and tutorials will be based on SONATA. For more information see the SONATA github page.

References

See the paper about BMTK: link.

Please cite BMTK as follows:

Dai et al. Brain Modeling ToolKit: An open source software suite for multiscale modeling of brain circuits. PLoS Comput Biol 16(11): e1008386. https://doi.org/10.1371/journal.pcbi.1008386

Acknowledgements

We wish to thank the Allen Institute for Brain Science founder, Paul G. Allen, for their vision, encouragement, and support.