search

dandi / dandi-archive

DANDI API server and Web app
https://dandiarchive.org
13 stars 12 forks source link

request: extract brain area metadata #1985

Open bendichter opened 3 months ago

bendichter commented 3 months ago

Based on usage feedback from NeuroDataReHack and from personal experience, searching for brain area is currently a bottleneck. @neurovium also included search by brain area in his search specification document.

Currently, brain area is required for some modalities, but it is buried. It is not currently extracted into asset or dandiset metadata, and requires reading individual NWB file contents. It is also in different places for different modalities.

Searching for brain area can therefore take a long time, particularly in datasets that have many assets. For the IBL dataset, we have been able to get this time down to 8:26 using LINDI (see discussion), but that's still not great. It would be much better to pre-extract this metadata, and provide it as asset and dandiset metadata, which would make searching much faster on the user side. It would also allow us to register terms against ontologies and controlled vocabularies and run analyses on the types of brain areas recorded.

Specifically:

  1. Modify the Dandiset and Asset metadata to include brain area
  2. Modify the CLI to extract this information from the NWB files
  3. Modify the NWB Inspector to promote including this information
  4. Modify the DLP to display this information
bendichter commented 3 months ago
  1. Modify the Dandiset and Asset metadata to include brain area

It looks like this is taken care of already with the "Anatomy" entries, which can be added under subject matter for both assets and dandisets. The (potentially) missing piece is that we want an UBERON (or other) uri. Most of these brain areas are Allen Institute Mouse CCFv3 abbreviations. Are there official URIs for these abbreviations? Is there a recommended way to map them to UBERON areas?

satra commented 3 months ago

pinging @lydiang on CCFv3 URIs