Pointers from our conversation

[yarikoptic]

• kerchunk
  • @yarikoptic might want to look either support for niftis could be added/needed
• https://github.com/neuroscales/nifti-zarr/pull/1/files -- attempt to store NIfTIs in zarr
• sidecar files for .nwb to overload the metadata -- was PR somewhere TODO
• descriptions.tsv (and .json to describe) file supported by BIDS to "summarize" _desc- entities used: https://bids-specification.readthedocs.io/en/stable/derivatives/common-data-types.html#descriptionstsv
• for aggregation across dandisets, we would probably like to follow smth like https://bids.neuroimaging.io/bep035 which ATM adds study- entity but I do not feel it makes sense in this case, let's go for ds- as abbreviation of dataset or dandiset ;-)
• ZEP0004 on ongoing effort to formalize "conventions" used within .zarr: https://github.com/zarr-developers/zarr-specs/pull/262/files
• possible extensions candidates: .lindi.zarr.json , .nwb.zarr.json, .nwb.lindi , .nwb.lindi.json, .lindi.json, .lindi

[magland]

Example lindi nwb json file

https://kerchunk.neurosift.org/dandi/dandisets/000232/assets/a040aa16-1091-4010-b171-9f136d880dd1/zarr.json