Information Processing Primitives

This is a reference implementation of the following paper:

Voges N., et. al. (2023) Decomposing neural circuit function into information processing primitives. Journal of Neuroscience

paper: https://www.jneurosci.org/content/early/2023/11/13/JNEUROSCI.0157-23.2023

Platform information and requirements

Platform: Ubuntu 22.04.2 LTS

Python: 3.9.16

Nest: 3.4

Matplotlib: 3.7.1

Seaborn: 0.12.2

SciPy: 1.10.1

numpy: 1.24.3

xarray: 2023.5.0

frites: 4.65.0

jupyter-lab: 3.3.0

conda: 23.3.1

Requirements installation

The easiest way to install the requirements needed to run the codes in this repository is to install them using anaconda (https://www.anaconda.com/download) and the enviroment file provided in this repository. Otherwise, the packages can be installed manually using pip or conda.

the commands bellow assumes that you are using the terminal in a linux system

First, clone this repository locally:

git clone https://github.com/brainets/IPP_PAPER

then, change to the clone directory:

cd IPP_PAPER

and install the packages via:

conda env create -f ipp_env.yml

Once the installation is finished, activate the enviroment

conda activate ipp

Now you should be able to run the codes in this repository.

FRITES installation

The easiest way to install FRITES (remember to activate the ipp conda env before) is via pip:

pip install frites

or

pip install git+https://github.com/brainets/frites

or

git clone https://github.com/brainets/frites
cd frites
python setup.py develop

For more information on FRITES check https://github.com/brainets/frites.

Repository content

This repository contains the code implementation for the ring model and the large-scale cortical model in the src directory:

Ring model

• integration.py: functions to perform numerical integration of the ring model.
• model.py: function to create the ring models and simulate them.
• utils.py: utility functions for the ring model (create simulation batches, stimulus kernels, etc).

Large scale model

• meanfieldcircuit.py: Implementation of the model using the Nest simulator https://www.nest-simulator.org/.
• setParams.py: Function to get parameters of the model.

Notebooks:

The notebooks can be found in the Notebooks subdirectory.

• "1. One Ring Model.ipynb: Notebook to simulate the one ring model."
• "2. Two Rings Model.ipynb: Notebook to simulate the two rings (2FF) model."
• "3. Three Rings Model.ipynb: Notebook to simulate the three rings (3FF) model."
• "4. DataSet.ipynb: Load the dataset from Markov et. al., (2013)."
• "5. Mean Field circuit IPP.ipynb: Load data for the large-scale circuit and produces figure 6 from the paper."

ATTENTION: In order to run notebook 5, one should first simulate the large-scale model, this can be done using the main.py script in the main directory:

python main.py 2

but this code can be memory consuming, and you might want to get a ''bigger'' computer before doing that!

Standalone C/C++ implementation

Copyright 2019 Johannes Hausmann, Nicole Voges ([email protected], [email protected])

The implementation of the analysis focused on the ring model was originally provided as a standalone C/C++ that we also make available at the src_C repository.

Inside this directory you will enconteur the following source codes:

• nmring.h/.cpp: main model, integration, and stimulus-related code.
• util.h/.cpp: utility macros and file system, averaging, PNG and numpy array writing methods.
• threadpool.h/.cpp: thread pool to run the model in a MT environment.
• oneRing_model.cpp: main driver program for a single ring.
• fb2rings_model.cpp: main driver program for two rings with feed-forward and -backward couplings.
• ff3rings_model.cpp main driver program for three rings with feed-forward coupling.

To use this version of the implementation first chang to the src_C directory:

cd src_C

and compile the code you want to execute. For compiling the code for all three models do:

make2

or a specific one (for instance the one ring configuration) by typing:

make oneRing_model

Clean compiled code by typing :

make clean 

To test functionality, comment in the lines at the beginning of each main function marked with "// minitest of functionality" (and comment out the lines marked with "// the real thing:") and compile.