I'm trying to import a BIDS folder from OMEGA. (Right click on protocol> Import BIDS dataset > Select folder to import
I receive the error:
** Error: [process_import_bids] Import > Import BIDS dataset
** Invalid BIDS dataset: missing file "dataset_description.json"
I have tried using various combinations of selecting folder levels, and have tried renaming the .json file found inside the subject's folder with the exact name: dataset_description.json. I am using an updated version of Brainstorm.
I think a mistake was made when packacing the BIDS folder for OMEGA where the dataset_description.json file was missing. @Marc.Lalancette can expand on that.
For the time being, you can create an empty dataset_description.json file at the root of your BIDS folder and the importer will not complain.
When you download OMEGA (which is temporarily not available at the moment due to ethics review unfortunately), there is indeed a dataset_description.json file in the root folder. It doesn't contain much: the dataset name and BIDS version. It should in the future be included when selecting subjects to download, but it can also be selected explicitly.
There was no such file in the main BIDS folder, but there were in the subject ones. Selecting them individually didn't solve the problem, but copying and renaming one of the subject level ones to the main BIDS folder did work, at least for the next step. It read in the subject but only created the head surface, and trying to get it to make brain surfaces resulted in a page full of errors and frozen Brainstorm. I will document these problems more fully soon.
I will be cognizant of your ongoing ethics review.
We have another question concerning OMEGA: In the questionnaires that are in the folder subject_metadata, there is a column named "timestamp". We noticed that the timestamp values in the questionnaires did not match with the acquisition time values found in the tsv files for a few subjects.
For example:
for sub-0030: the dates don't match (year 2015 in the Omega general screening questionnaire and year 2013 in the tsv files for the two available sessions).
for sub-0034, sub-0041 and a few other subjects dates match and time differs slightly.
Could you please let us know what these timestamp values correspond to? This is important as we need to know the time of the recordings.
As Martin said, just creating an empty file named dataset_description.json in the right place would also have worked. The ..._meg.json files next to the MEG data contain other recording-specific metadata.
The questionnaire timestamp is when the questionnaire was filled. It's normal that it's not always the same day as the MEG recordings sometimes. In fact that's why there's a "retrospective" version, with fewer questions that would only be relevant at the time of the scan. Note that many participants are lab members, so it was possible to have them fill the questionnaire much later than the scans, when the questionnaires were first implemented or a new version introduced. That's also why a few participants appear in multiple spreadsheets.
However, there were a few subjects numbers that were re-used after some data was discarded from the archive. I recently fixed a couple of these that had the wrong questionnaire data. I can send you the latest files privately. But specifically, the subjects you mentioned are ok.