Participants.tsv in BIDS: Recommended, Not Required

[Peyman-N]

We (myself and @spieschnik) were checking the bids-example. We encountered an issue while running bids2openminds on the asl003 dataset. Upon further investigation, it appears that the participants.tsv file is not required. As a temporary solution, I propose modifying the code to skip subject creation if the file is not present. Additionally, in the future, I suggest implementing a backup method based on directory patterns to extract this information.

[lzehl]

@Peyman-N can you please run asl003 through a BIDS validation to see/share the output of the validation please?

[lzehl]

Modality agnostic files (not all) according to the BIDS specifications (human readable one):

• dataset_description.json -> REQUIRED
• README(.md/.rst/.txt) -> REQUIRED (MUST be in ASCII or UTF-8 encoding)
• CITATION.cff -> RECOMMENDED (MUST follow CFF specs)
• CHANGES -> RECOMMENDED (MUST follow CPAN::Changes::Spec)
• LICENSE -> RECOMMENDED (MUST be in ASCII or UTF-8 encoding)
• participants.tsv, participants.json -> RECOMMENDED
• samples.tsv, samples.json -> REQUIRED (if sample-<label> is present)

[lzehl]

We can discuss if we actually push people to a MUST participant.tsc/json when they want to convert to openMINDS.

[Peyman-N]

@Peyman-N can you please run asl003 through a BIDS validation to see/share the output of the validation please?

@lzehl Here is the results of BIDS validator. It seems that validator don't care about participants files.

[apdavison]

We can discuss if we actually push people to a MUST participant.tsc/json when they want to convert to openMINDS.

I think we should allow conversion to openMINDS for such cases, but we may not accept such datasets in EBRAINS.

[Peyman-N]

We can discuss if we actually push people to a MUST participant.tsc/json when they want to convert to openMINDS.

I think with current structure of openMINDS and its emphasis on subject and subject states, we are implicitly guiding (nudging) them to do so.

[lzehl]

@apdavison @Peyman-N my point is rather: if this file is not given the converter will not provide any information on subjects (also not inferred from the file names). And the converter could provide a report/assessment for the user at the end with errors, strong suggestions, and optional additions.

[lzehl]

@lzehl Here is the results of BIDS validator. It seems that validator don't care about participants files.

@Peyman-N which fits what I extracted from the human readable BIDS specs. But since validation fails for this example dataset: this also means we should test all example datasets we use to know their behavior in a BIDS validator.