Skip to content

Neurobagel graph data files


Using the Neurobagel CLI (see also the section on the CLI), a Neurobagel data dictionary (.json) for a dataset can be processed together with the corresponding tabular data (.tsv) and BIDS dataset (if available) to generate subject-level linked data that can be encoded in a knowledge graph. The Neurobagel graph-ready data are stored in JSON-LD format (.jsonld), and include a representation of each subject's harmonized phenotypic properties and imaging metadata.

Another way to think about the difference between a Neurobagel data dictionary and a graph-ready .jsonld data file is this: more than one dataset can theoretically have the same data dictionary (if the tabular data include the same columns and unique column values), but the .jsonld file for each dataset is unique as long as the actual data of subjects differs across datasets.

Example .jsonld files

Depending on whether a dataset annotated using Neurobagel includes BIDS imaging data, the .jsonld data for the dataset may or may not include imaging metadata of subjects (extracted automatically with the CLI).

More info on example dataset

The above .jsonld files represent an example dataset used for testing which includes the following:

Data Link
Phenotypic TSV
Neurobagel data dictionary
BIDS dataset