Automated conversion of BIDS datasets to Linked Data using openMINDS
Poster No:
2252
Submission Type:
Abstract Submission
Authors:
Sophia Pieschnik1, Peyman Najafi2, Jan Bjaalie1, Trygve Leergaard1, Andrew Davison2, Lyuba Zehl3
Institutions:
1University of Oslo, Oslo, Norway, 2Université Paris-Saclay, CNRS, Saclay, France, 3EBRAINS AISBL, Brussels, Belgium
First Author:
Co-Author(s):
Introduction:
BIDS [1] and openMINDS [2] are expert- and community-driven standards that facilitate FAIR data management in the field of neuroscience [3].
BIDS is a community-wide accepted data model that standardises the file organisation across multiple human neuroscience modalities. It includes restrictions for formatting data and metadata as well as strict naming conventions for folders and files [4]. Although the hierarchical structure is easy to read and interpret by humans, it lacks explicit links between data and metadata. As a result, tools have to hard code the implicit relations of data and metadata, indicated through the naming convention and enumerated types.
The openMINDS metadata framework provides an extendable set of (i) integrated metadata models for Linked Data and (ii) libraries of well-defined, ontology-based metadata instances [5]. Both components enforce an explicit linkage between metadata, data, and external resources such as ontologies. The local file organisation is unconstrained by design, to guarantee maximum flexibility.
BIDS and openMINDS are complementary. The former ensures that metadata are always available together with downloaded data files, while the latter supports rich search capabilities in data repositories based on Linked Data, such as EBRAINS. To facilitate the ingestion of BIDS-formatted datasets into such repositories, we have developed an easily-installable command-line tool bids2openminds.
BIDS is a community-wide accepted data model that standardises the file organisation across multiple human neuroscience modalities. It includes restrictions for formatting data and metadata as well as strict naming conventions for folders and files [4]. Although the hierarchical structure is easy to read and interpret by humans, it lacks explicit links between data and metadata. As a result, tools have to hard code the implicit relations of data and metadata, indicated through the naming convention and enumerated types.
The openMINDS metadata framework provides an extendable set of (i) integrated metadata models for Linked Data and (ii) libraries of well-defined, ontology-based metadata instances [5]. Both components enforce an explicit linkage between metadata, data, and external resources such as ontologies. The local file organisation is unconstrained by design, to guarantee maximum flexibility.
BIDS and openMINDS are complementary. The former ensures that metadata are always available together with downloaded data files, while the latter supports rich search capabilities in data repositories based on Linked Data, such as EBRAINS. To facilitate the ingestion of BIDS-formatted datasets into such repositories, we have developed an easily-installable command-line tool bids2openminds.
Methods:
bids2openminds is an open-source project on GitHub under the MIT License [6].
Both standards, BIDS and openMINDS, are continuously extended to cover more areas of neuroscience. As a starting point bids2openminds supports the BIDS modalities structural and functional Magnetic Resonance Imaging (MRI), Electroencephalography (EEG), and Intracranial Electroencephalography (iEEG).
The converter is provided as a Python package, building on the functionalities of two existing Python packages, pyBIDS and openMINDS_Python. pyBIDS allows handling of BIDS data and metadata while openMINDS_Python enables the creation and manipulation of openMINDS-compliant metadata collections.
At present the semantic mapping from BIDS to openMINDS is hard coded, since BIDS does not currently provide unique identifiers for the schemata, properties, and enumerated types, but in future this could be auto-generated.
Both standards, BIDS and openMINDS, are continuously extended to cover more areas of neuroscience. As a starting point bids2openminds supports the BIDS modalities structural and functional Magnetic Resonance Imaging (MRI), Electroencephalography (EEG), and Intracranial Electroencephalography (iEEG).
The converter is provided as a Python package, building on the functionalities of two existing Python packages, pyBIDS and openMINDS_Python. pyBIDS allows handling of BIDS data and metadata while openMINDS_Python enables the creation and manipulation of openMINDS-compliant metadata collections.
At present the semantic mapping from BIDS to openMINDS is hard coded, since BIDS does not currently provide unique identifiers for the schemata, properties, and enumerated types, but in future this could be auto-generated.
Results:
With bids2openminds we present a command-line tool and Python package to the neuroscience community that converts the metadata from a BIDS compliant dataset into openMINDS (as a collection of JSON-LD documents), in order to represent BIDS data in graph databases. Note that openMINDS is a continuously growing metadata model, which does not yet provide extensions for all neuroscience modalities integrated in BIDS.
We demonstrate the use of bids2openminds for data published on the bids-examples GitHub repository [7] as outlined in Figure 1. We used the resulting openMINDS-compliant metadata collections to register these data in EBRAINS, providing insight into the advantages of Linked Data integrations: The registration enables users to search, filter and explore the data and metadata via the search user interface and via programmatic queries. Moreover, data are also linked to other databases by tagging research products with openMINDS ontology-driven controlled terminologies.
We demonstrate the use of bids2openminds for data published on the bids-examples GitHub repository [7] as outlined in Figure 1. We used the resulting openMINDS-compliant metadata collections to register these data in EBRAINS, providing insight into the advantages of Linked Data integrations: The registration enables users to search, filter and explore the data and metadata via the search user interface and via programmatic queries. Moreover, data are also linked to other databases by tagging research products with openMINDS ontology-driven controlled terminologies.
·Outline of bids2openminds functionality
Conclusions:
bids2openminds enables an easier registration of BIDS compliant datasets in openMINDS based graph databases, and thus further automates data curation workflows. Such explicitly linked graph databases offer advanced query options for heterogeneous neuroimaging data, promoting findability, and enable further annotations and cross-links going beyond the BIDS data model, to facilitate interoperability and reuse. Thus, bids2openminds promotes FAIR data sharing for neuroscience data.
Neuroinformatics and Data Sharing:
Databasing and Data Sharing 2
Informatics Other 1
Keywords:
Data Organization
Data Registration
Informatics
Open-Source Code
1|2Indicates the priority used for review
Provide references using author date format
2. Open Metadata Initiative (2020) openMINDS metadata framework. RRID: https://scicrunch.org/resolver/SCR_023173. URL: https://openminds.openmetadatainitiative.org/
3. Wilkinson, M. et al. (2016) The FAIR Guiding Principles for scientific data management and stewardship. Scientific Data 3 ,160018), DOI: 10.1038/sdata.2016.18
4. BIDS specification (2023, Nov 30) Read the Docs. URL: https://bids-specification.readthedocs.io/en/stable/
5. openMINDS documentation (2023, Nov 30) Read the Docs. URL: https://openminds-documentation.readthedocs.io/en/latest/
6. bids2openminds (2023, Dec 1) GitHub. URL: https://github.com/openMetadataInitiative/bids2openminds
7. bids-examples: eeg_rest_fmri (2023, Nov 11) GitHub. URL: https://github.com/bids-standard/bids-examples/tree/master/eeg_rest_fmri