[Feature Request] Automatically add used ontologies to the investigation.

[Thyra]

Describe the solution you'd like
When filling in cells in a study or assay table, you automatically get suggestions for ontology terms but currently have to manually go back to the investigation and add the ontologies you used there (and it's not even that easy because when adding the term through the search it doesn't tell you the ontology version or URL used). I think most users won't do this (especially if they're not comfortable with ontologies) and at the same time it's very straightforward to automate, at least conceptually. So it'd be a nice added value!

Additional context
I wasn't sure if this repo is the right place for it or swate, decided for this one because it concerns multiple files at the same time and that's not how swate is used always (I think? e.g. https://swate-alpha.nfdi4plants.org/ ). LMK if I'm wrong here and should transfer the issue.

[Freymaurer]

It is totally fine to open these issues on Swate or ARCitect, we will try to sort them whereever we think they belong 😄

This feature requires actually interop between ARCitect and Swate and this should be based on a ARCtrl function

I will move this issue first to ARCtrl to discuss how this could be implemented and when resolved on that end we will move it to Swate.

@HLWeil What do you think would be required to have a function on ArcInvestigation level to collect all used ontologies. What information do you need for this?

[muehlhaus]

There is no need to explicitly list the ontologies within the ARC Investigation file, as this information is automatically aggregated during the creation of the ARC RO-Crate. Instead, the ARC structure should prioritize file-level separation of concerns to support meaningful and manageable version control. For example, a change to an assay file should not necessitate a corresponding update to the investigation file. If multiple files always need to be edited in tandem, this undermines the goal of modular and traceable commits.

[Freymaurer]

What is the intented best practise of this approach and the MUST contain entry for Ontology Source References in isa investigation? (https://github.com/nfdi4plants/ARC-specification/blob/release/ISA-XLSX.md#ontology-source-reference)

[Thyra]

Interesting point! (although I think the modularity of commits is somewhat limited anyhow as an unfortunate side effect of the binary XLSX format). But then as @Freymaurer alludes to, I think it should be taken out of the interface. Right now we have the worst of both worlds ;-)

[HLWeil]

I think @muehlhaus brought a good argument against updating this section every time you add an ontology term somewhere. Additionally, I would add that the information contained in this section is usually not of much value anyways, as many ontologies don't really have proper versions and the terms placed in the Annotation tables also contain the URL.

But, after a discussion with @Thyra we identified a potential use-case for this section, that was probably also one of the main use cases for the original specification in ISA(-Tab): Custom/Local Ontologies

If a user wants to add Ontology Terms not easily accessible online it might be a middle of the way solution to ship the ontology with the ARC and point to it from the Ontology Term and the Ontology Source References section. If we want to support this case, we would have to keep this section, but would not really need any automatic filling. It would be a manual field.

My main reasoning for having this section without filling it automatically is the parsing from and to ISA-json. Even if we don't use this information, we can still support a lossless represantion of ISA-Json in the ARC Scaffold.