[Discussion]: When does it make sense to not use NWB?

[bendichter]

What would you like to see added to the NWBInspector?

I wanted to bring up a topic that came up with @pauladkisson regarding when it might be a good idea to store data outside of NWB and outside of DANDI.

Paul is working on a dataset conversion where the lab uses Western blots, which results in data that are TIF images. Of course it's not hard to add these to the NWB file as Image objects, but these blots were run on a separate set of animals, so it would not be straightforward to attach them to NWB files corresponding to specific neurophysiology sessions.

This type of situation has come up before. With the Buffalo Lab, their task used an environment they set up in Unity. Developing and extension and storing that in NWB would be quite difficult and likely out of scope for a data conversion.

In both of these cases, it seems like it might be better to store data in an alternative archive and link to this dataset as a related resource in a Dandiset.

This brings up a lot of questions for applying this logic more broadly:

• Under what circumstances would you want to store data outside of NWB and outside of DANDI? When would it be better to use an extension structure like LabMetaData?
• Where should such data go? What types of requirements should we have for external data sources?
• What types of files and data representations should we allow?
• How do you associate the NWB data with this other dataset?

Do you have any interest in helping implement the feature?

Yes.

Code of Conduct

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