ebids

Code to translate electrophysiological data to BIDS format.

Installation

eBIDS

Currently macOS/Linux only.

Add the ebids binary directory to your path. For example, if ebids is in your home directory under analysis:

export PATH=$PATH:$HOME/analysis/ebids/bin

Also add ebids to your python path:

export PYTHONPATH=$PYTHONPATH:$HOME/analysis/ebids

To install the Matlab scripts, add ebids/matlab to your Matlab path.

To run those commands automatically, put them in $HOME/.bashrc.

Dependencies

To work with Neuralynx data in python (used for some preprocessing scripts), download our fork of the Neo project and install by adding the project directory to your PYTHONPATH (similar to above for ebids).

If using python for analysis, you'll also want pybids: pip install pybids. If using Matlab, try bids-matlab.

Converting task and recording data

See the wiki for information about converting specific tasks and recording formats. This project is intended to collect scripts for converting multiple tasks, to provide multiple examples and avoid solving the same problems multiple times. However, to keep the project focused, it will only include scripts for converting data, and possibly for loading BIDS data, but not for analysis.

Syncing task data with recordings

See alignment for instructions for aligning TTL pulses to sync up task data with recording data.

Reading BIDS data

For BIDS compatibility, in addition to running the relevant data conversion scripts, you also need dataset_description.json and participants.tsv files in the bids_dir.

Python

In python, you should then be able to validate the BIDS formatting and load information about the dataset. At this point, you can easily do things like load behavioral data for a task run.

from bids import BIDSLayout
layout = BIDSLayout(bids_dir) # validate dataset and load information
data = layout.get(task='xmaze', subject='DS1902', session='01', run=1).get_df() # load run events
data.loc[data.trial_type == 'reward',:] # get just the reward events

MATLAB

Matlab support seems to be less well-developed, but also works:

layout = bids.layout(bids_dir); % load dataset information
data = struct2table(layout.subjects(1).func(1).meta); % load run events