CxSystem is a cerebral cortex simulation framework, which operates on personal computers. We have tested the simulation software with a simplified version of a comprehensive cortical microcircuit model [1], which is available as configuration file. The CxSystem aims at easy testing and buildup of diverse models at single cell resolution. Implemented on the top of the Python-based Brian2 simulator, CxSystem supports the main goal of Brian, i.e. minimizing development time, by providing the user with a simplified interface. The preliminary version of this software has been developed at Aalto University 2012-2015 [2], and the full version at the University of Helsinki 2013-2017. The software is distributed under the terms of the GNU General Public License, Copyright 2017 Vafa Andalibi, Henri Hokkanen and Simo Vanni. To access the documentation of the CxSystem, please check out the readthedocs page of the project.

---

[1] Markram, H., Muller, E., Ramaswamy, S., Reimann, M. W., Abdellah, M., Sanchez, C. A., … Schürmann, F. (2015). Reconstruction and Simulation of Neocortical Microcircuitry. Cell, 163(2), 456–492. http://doi.org/10.1016/j.cell.2015.09.029
[2]Heikkinen, H., Sharifian, F., Vigário, R., & Vanni, S. (2015). Feedback to distal dendrites links fMRI signals to neural receptive fields in a spiking network model of the visual cortex. Journal of Neurophysiology, 114(1), 57–69. http://doi.org/10.1152/jn.00169.2015

The following examples are currently available for simulation:

We have prepared several configurations files for sample networks of different examples:

Pytest

Click here to load the configuration for the pytest example which is basically a generic sample network.

COBA-EIF

Click here to load the configuration for the COBA-EIF example.

Simplified Markram Model (step 2)

This model is a simplification of the Markram et al. (2015) model of the rat somatosensory cortex containing ~31 000 neurons divided into 17 neuron groups. Details of the simplification are described in the article. Running this model should take ~6 minutes per 1 second of simulation time (device=Python).

Click here to load the configuration of the step 2 simplification.

Simplified Markram Model (step 1)

This model contains all the 55 neuron groups of the Markram et al. (2015) rat somatosensory cortex model. The neuron dynamics have been reduced to exponential integrate-and-fire. Running this model should take ~17 minutes per 1 second of simulation time (device=Python).

Click here to load the configuration of the step 1 simplification. Your browser might become unresponsive while the configuration is being loaded. Be patient.