Credit Assignment Project Code

1. Description

This repository contains the code for analyzing the data from the Credit Assignment project, an Allen Institute for Brain Science OpenScope project.

The experiment details, analyses and results are published in Gillon et al., 2021, bioRxiv.

2. Installation

To run the code, you should install Anaconda or Miniconda.

Once these are installed, you can simply use the appropriate .yml file to create a conda environment. For example, if using Ubuntu or Mac OS, open a terminal, go to the repository directory, and enter:

1. conda env create -f osca.yml
2. source activate osca

The code is written in Python 3.

3. Use

Once installed, when using the codebase, simply activate the environment:

source activate osca

All of the appropriate libraries should then be loaded, and the modules can be imported for use in ipython, python scripts, or jupyter notebooks, for example.

4. Scripts and modules

• run_paper_figures.py: run, analyse and plot paper figures (for example usage, see the paper_figures folder)
• analysis: analysis scripts, including the Session and Stim objects
• sess_util: session specific utilities module
• plot_fcts: plotting scripts
• paper_fig_util: scripts to organize and generate the paper figures
• examples: example notebook for using the Session and Stim objects

5. Data

The full dataset for this project is hosted here in the DANDI archive in NWB format. The associated metadata can be found here. The subset of data used in the paper (33 sessions, ~15 GB total) can be downloaded by running, from the main directory of the repository:
python sess_util/sess_download_util.py --output path/to/save/

Code to generate the stimuli used in these experiments can be found here.

6. Example notebooks

The following notebooks give examples of how to download the data, and run the paper analyses.

Run in Binder	View the notebook	Run the Google Colab notebook

The contents of the Binder and Google Colab notebooks differ somewhat, due to the resources available:

• Binder: conda env. is already installed (+), but only limited compute resources are available (-).
• Google Colab: conda env. must first be installed (-), but more substantial compute resources are available (+).

7. Authors

This code was written by:

• Colleen Gillon (colleen dot gillon at mail dot utoronto dot ca)
• Jay Pina, Joel Zylberberg, and Blake Richards

Please do not hesitate to contact the authors or open an issue/pull request, if you have trouble using the data or the codebase or improvements to propose.

Note: The module Dataset2p.py under sess_util contains code shared by authors at the Allen Institute for Brain Science. The authors of the code cannot guarantee support for its usage.