Bech & Dard NWB process repository

NWB files processing

This repository allows the processing NWBs files to reproduce analysis from Bech & Dard, eLife 2026.
This README provides instruction to set up the python environment and run the analysis code to generate intermediate data used to plot.
See at the end how to reproduce the figure panels starting from the intermediate dataset. (Bech, Dard figures repo)

Installation

Create environment

conda create -n bech_dard_nwb_process python=3.11
conda activate bech_dard_nwb_process 

Install git if necessary

conda install git

Install dependencies

cd /path/to/Bech_Dard_process_NWB
pip install -e .

How to use - Reproduce intermediate data

Run each file sequentially

conda activate bech_dard_nwb_process 
python path/to/repo/main_analysis/figure1_analysis.py
python path/to/repo/main_analysis/figure1_supp_analysis.py
python path/to/repo/main_analysis/figure2_analysis.py
python path/to/repo/main_analysis/figure2_supp_analysis.py
python path/to/repo/main_analysis/figure2_supp_stats.py
python path/to/repo/main_analysis/figure3_analysis.py
python path/to/repo/main_analysis/figure3_supp_analysis.py
python path/to/repo/main_analysis/figure4_analysis.py
python path/to/repo/main_analysis/process_deeplabcutdata.py
python path/to/repo/main_analysis/process_opto_widefield_examples.py
python path/to/main_analysis/model_context_behaviour.py
python path/to/repo/main_analysis/pixel_correlation_analysis.py
python path/to/repo/main_analysis/pixel_correlation_processing.py

Warning : pixel_correlation_analysis was previously optimized to run on HPC.
To run a test change in the pixel_correlation_analysis.py file the number of shuffles from 1000 to 20

Output

Each script is going to populate a results folder created within the main folder.

Reproduction of figures panels

Reorganize results to follow figure panel ordering

conda activate bech_dard_nwb_process 
python path/to/repo/main_analysis/panel_data_format.py

This will create a 'published_data' folder within the main result folder.
This matches publicly available data on Zenodo

Figure making

To reproduce the figure panels from the downloaded or generated intermediate dataset see:

• Bech, Dard figures repo
• Intermediate dataset