This is a collection of functions for managing the experimental data recorded in the BeNeuro Lab, and a CLI tool called bnd for easy access to this functionality.

Features so far:

• validating raw experimental data on the computers doing the recording
• uploading experimental data to the RDS server
• downloading experimental data from the RDS server
• running spike sorting

Features on the way:

• checking and uploading automatically on a schedule
• converting processed data to NWB
• converting NWB to TrialData

1. Create an empty conda environment in which you will install the tool. The environment only needs Python and pip. You can do this with mamba create -n bnd python pip.

   Alternatively you can install with the system Python, but that's not really recommended. If you have conda/mamba on the computer, just use that for a peace of mind for now.

   You will also need poetry. If you don't already have it installed, follow one of installation methods here, which on Linux will most likely be one of the following:

   pipx install poetry

   curl -sSL https://install.python-poetry.org | python3 -

   (Note that it is not advised to install poetry in the environment you just created.)

2. Clone this repo

   git clone https://github.com/BeNeuroLab/beneuro_experimental_data_organization.git

3. Navigate into the folder you just downloaded (beneuro_experimental_data_organization)

4. Activate the environment you installed in.

   conda activate bnd

5. Install the package with either

   poetry install

   or if you want spike sorting functionality:

   poetry install --with processing

   For more info, see the spike sorting instructions.

6. Test that the install worked with

   poetry run pytest

   Hopefully you'll see green on the bottom (some yellow is fine) meaning that all tests pass :)

The tool needs to know where the experimental data is stored locally and remotely.

0. Mount the RDS server. (If you're able to access the data on it from the file browser, it's probably already mounted.)

1. Run bnd init and enter the root folders where the experimental data are stored on the local computer and the server. These refer to the folders where you have raw and processed folders.

   This will create a file called .env in the beneuro_experimental_data_organization folder and add the following content:

   LOCAL_PATH = /path/to/the/root/of/the/experimental/data/storage/on/the/local/computer
   REMOTE_PATH = /path/to/the/root/of/the/experimental/data/storage/where/you/mounted/RDS/to
   

   Alternatively, you can create this file by hand.

2. Run bnd check-config to verify that the folders in the config have the expected raw and processed folders within them.

• To see the available commands: bnd --help
• To see the help of a command (e.g. rename-videos): bnd rename-videos --help

• You can validate the structure of raw data for an individual session:

  • bnd validate-session . <subject-name> if you're in the session's directory
  • bnd validate-session /absolute/path/to/session/folder <subject-name> from anywhere
  • bnd validate-last <subject-name> from anywhere to validate the last recorded session
  • bnd validate-today from anywhere to validate all recorded sessions on the current day
    • If it's trying to validate things in places like "treadmill-calibration" that are on the same level as subject directories, you can exclude checking in those places by adding them to IGNORED_SUBJECT_LEVEL_DIRS in the .env config file (IGNORED_SUBJECT_LEVEL_DIRS = ["treadmill-calibration", "other-stuff-you-want-to-ignore"])
    • bnd list-today lets you check what sessions were recorded on the current day

  This will give you an error if there is a problem with the file structure.

  The name of the subject is used for confirmation, but might be removed in the future if it's too annoying.

• or for all sessions of a subject:

  • bnd validate-sessions <subject-name>

  This will give you an overview which sessions look good and which ones have a problem.

By default behavioral, ephys, and video data are all checked. To control which kind of data you want to check:

• To exclude checking something: --ignore-behavior, --ignore-ephys, --ignore-videos
• To explicitly include something: --check-behavior, --check-ephys, --check-videos

Please note that running validation will only give you the first problem that pops up. Once you fixed that, run it again to see if there are others ;)

The default naming Jarvis uses for the video folder and files doesn't match the convention we want to follow.

Files can be renamed with bnd rename-videos . <subject-name> (or specifying the path instead of . if the current working directory is not the session's directory).

Add --verbose to the end to see what files were renamed.

Sometimes the experimenter leaves comments in a comment.txt file or saves some extra .txt files in the electrophysiology recording folders.

To rename these files to follow the naming convention of <session-name>_<filename>, you can use the bnd rename-extra-files command.

Once you're done recording a session, you can upload that session to the server with:

bnd upload-session . <subject-name>

or if you don't want to cd into the session's directory:

bnd upload-last <subject-name>

This should first rename the videos and extra files (unless otherwise specified), validate the data, then copy it to the server, and complain if it's already there.

Downloading data to your local computer is similar to uploading, but instead of throwing errors, missing or invalid data is handled by skipping it and warning about it.

Using the session's path, e.g. after navigating to the session's folder on RDS mounted to your computer:

bnd download-session . <subject-name>

or just the last session of a subject:

bnd download-last <subject-name>

Currently we are using Kilosort 4 for spike sorting, and provide a command to run sorting on a session and save the results in the processed folder.

Note that you will need some extra dependencies that are not installed by default, and that this will most likely only work on Linux.
You can install the spike sorting dependencies by running poetry install --with processing in bnd's root folder.

You will also need docker to run the pre-packaged Kilosort docker images and the nvidia-container-toolkit to allow those images to use the GPU.
If not already installed, install docker following the instructions on this page, then install nvidia-container-toolkit following these instructions.

Basic usage:

bnd kilosort-session . M020

Only sorting specific probes:

bnd kilosort-session . M020 imec0

bnd kilosort-session . M020 imec0 imec1

Keeping binary files useful for Phy:

bnd kilosort-session . M020 --keep-temp-files

Suppressing output:

bnd kilosort-session . M020 --no-verbose