HED test datasets¶
This repository contains a set of HED-annotated datasets in BIDS-compatible format. These datasets can be useful for:
Writing lightweight software tests.
Serving as examples of how to incorporate HED into BIDS-structured data.
The datasets have empty raw data files. However, some data headers containing the metadata are still intact.
Datasets that are derived from datasets on OpenNeuro are identified by their OpenNeuro accession number plus ‘s’ plus a modifier. Datasets focused on particular a particular modality may have the modality prepended to the name. For example, eeg_ds003645s identifies a reduced dataset derived the EEG data in OpenNeuro dataset ds003645. The suffix modifier indicates what this dataset is designed to test.
Short-form tags with definitions.
Multiple sidecars with inheritance.
Multiple HED library schemas.
Long-form with definitions.
Annotation with single column.
This dataset was originally released as a multi-modal dataset by
Detailed case study in using HED for tagging:
Robbins, K., Truong, D., Appelhoff, S., Delorme, A., & Makeig, S. (2021, May 7). Capturing the nature of events and event context using Hierarchical Event Descriptors (HED). BioRxiv, 2021.05.06.442841. https://doi.org/10.1101/2021.05.06.442841
This is eeg_ds003645s_hed_inheritance
This dataset is designed to test the HED library schema facility. It uses HED 8.0.0 as a base schema and as the “test” library schema. In addition, this dataset uses the SCORE library version 1.0.0 as a library schema.
The schema used are specified in the
dataset_description.json file using the
This dataset is used to test that tools work with either long-form or short-form HED tags.
This dataset is designed to illustrate a basic FMRI pipeline.
For general information on the
bids-validator, including installation, configuration, and usage,
see the bids-validator README file.
Example: The following command validates the
bids-validator eeg_ds003645s_hed --config.ignore=99
This example assumes that
npm and the
bids-validator npm package
have been installed on the local machine.
The command is run from the directory above the dataset root directory.
--config.ignore=99 flag tells the bids-validator to ignore empty data files
rather than to report the empty file error.
For additional information on BIDS validation, see the bids-examples.