[Feature]: Detect when an NWB file has no domain-specific neurodata types

[bendichter]

What would you like to see added to the NWBInspector?

Occasionally, we get NWB file submissions on DANDI that contain all of their data in TimeSeries objects. These datasets generally contain doman-specific data that should use e.g. SpatialSeries and ElectricalSeries, but use the generic type. This means the files don't have the associated metadata and can't be correctly categorized by DANDI.

See e.g. dandi/helpdesk#159

Is there a way of detecting and flagging this?

(request from @satra)

Do you have any interest in helping implement the feature?

No.

Code of Conduct

• I agree to follow this project's Code of Conduct
• Have you ensured this feature was not already requested?

[CodyCBakerPhD]

The basic principle of asserting a certain complexity/diversity of file contents is appealing, but determining good thresholds for how to reliably detect it might be tricky

Another consideration is to avoid false positives on 'split' file strategies such as Spyglass, where multiple NWB files sharing the same session IDs / start times constitute an 'NWB folder' (a viewing strategy facilitated by Neurosift), and where each file may only have a small number of contents

It would help if we could find any other similar examples on DANDI to help craft some specific assertions common across the datasets

Otherwise, like Satra said on Slack, this seems like something more appropriate for curation by the scientific core combined with outreach to help them make the file better