SPADE_surrogates

Repository containing code necessary to reproduce the results of Stella, A., Bouss, P., Palm, G., & Grün, S. (2021). Generating surrogates for significance estimation of spatio-temporal spike patterns

Cloning the repo

Please clone the repository by running:

git clone --recurse-submodules
[email protected]:INM-6/SPADE_surrogates.git

Alternatively, you can proceed by:

git clone [email protected]:INM-6/SPADE_surrogates.git
git submodule update --init

In order to download the data, please, go into the multielectrode_grasp folder by typing:

cd data/multielectrode_grasp/datasets

Continue with

gin login

Possibly, you will need to run

gin init

if you didn't you use gin before on your computer.

If you don't have a gin-account or client, refer to the README in the multielectrode_grasp folder.

Checkout the release 1.1.0 of that repo by running:

git checkout 1.1.0

Then you need to run:

gin get-content i140703-001-03.nev l101210-001-02.nev i140703-001.odml l101210-001.odml

This will download the results of the spike-sorting (.nev) and metadata files (.odml).

If this fails to download the files, you can try git-annex as an alternative:

git-annex get i140703-001-03.nev l101210-001-02.nev i140703-001.odml l101210-001.odml

Creating the conda environment

For general instructions on how to use the conda environments, please refer to: https://conda.io/projects/conda/en/latest/user-guide/tasks/manage-environments.html.

To install the environment, first check that you have libopenmpi-dev installed. If not, run (for Linux):

sudo apt install libopenmpi-dev

Continue by installing the environment with:

conda env create -f env.yml
conda activate surrogates
pip install -e .

Creating the figures

Please go to the code folder cd SPADE_surrogates

• For Figure 3:
  • python fig_r2gexperiment.py
• For Figure 4:
  • python fig_surrogate_statistics_data.py
  • python fig_surrogate_statistics_plot.py
• For Figure 5 & Figure 6C:
  • python generate_original_concatenated_data.py
  • python fig_spikeloss_r2gstats.py
• For Figure 6B:
  • python fig_analytical_spike_loss.py
• For Figure 6C:
  • you get it with Figure 5
• For Figure 7:
  • cd fig_pvalue_spectrum
  • python max_order_statistics.py
  • python create_independent_spiketrains.py
  • mpirun python analyze_independent_spiketrains.py (This step should be done on a cluster)
  • python plot_pvalue_spectrum.py
  • cd ..
• For Figure 8:
  • cd analysis_artifial_data
  • snakemake (This step should be done a cluster.)
  • cd ..
  • python fig_artificial_data.py
• For Figure 9:
  • run the scripts for Figure 8
  • python fig_fps_fr_cv.py
• For Figure 10:
  • python generate_original_concatenated_data.py
  • cd analysis_experimental_data.py
  • snakemake (This step should be done a cluster.)
  • cd ..
  • python fig_experimental_data.py
• For Figure 11:
  • download and unzip https://borgelt.net/src/pycoco.zip
  • in pycoco directory run pip install -e .
  • python S2_fig_coconad.py

Scripts for Figures 3, 4, 5, 6, 11 can be run locally on a laptop. Some may take ~5-10 minutes in runtime, depending on the complexity of the analysis. Data generation and analysis steps that for computation require a cluster are indicated in brackets.