Use LinkML to manage RO-Crate archives

[glyg]

I would like to use LinkML to define and manage RO-Crate archives

Describe the solution you'd like

I would like to define the RO-Crate schema in LinkML so that I can extend it to my use case (microscopy data) and benefit from the linkML tooling to automate import/export workflows

How important is this feature? Select from the options below:
• Medium - can do work without it; but it's important (e.g. to save time or for convenience)

When will use cases depending on this become relevant? Select from the options below:
• Mid-term - 2-4 months

Additional context

I also opened an issue on the RO-Crate org here with more details: ResearchObject/ro-crate#264

It seems to me that both tools could benefit from each other. I can dedicate some time to implement a workflow, but would enjoy some pointers and feedback

Thanks!

Guillaume

[cmungall]

Hi @glyg, I think it would be great to do this and we'd be happy to work with you on this!

I think the first thing to establish is whether the linkml schema should model:

• the expanded JSON-LD structure
• the RO-crate data model

This may seem a bit esoteric distinction but starting on the right foot here is the best way to avoid any confusion. I love JSON-LD but it does sometimes create a bit of confusion. I am seeing more cases now of expanded JSON-LD being the form that is expected by tools which IMO seems a bit contrary to the original aims.

Some background on this:

• https://linkml.io/linkml/howtos/using-jsonld.html
• Add how-to guide for working with schema.org json-ld files, and for science profiles of schema.org #1385
• validation issues #532

My approach would be (2), to have the LinkML model reflect the RO-crate data model rather than trying to model fields like @id. This means that (a) the generated jsonschema won't be able to directly validate the expanded form (b) there may be a bit of additional tooling needed to generate the correct expanded form from python objects.

Let me know if that makes sense. And I am not an expert on RO-crates, cc-ing @stain to check my understanding!

[glyg]

Thanks a lot for the feedback, I still have some work to do to understand all the nooks and cranes here, but will come back with some concrete work soon :)

[cmungall]

Just a friendly ping to see if you need any further clarification!

[glyg]

Hi @cmungall thank you for your message! I don't need clarifications right now, I am working on a concrete case right now (not full time though). I should report soon-ish here!

[glyg]

Hi, sorry for the stale issue. I posted a brief review of my attempt on forum.image.sc

It's OK for me to close this if you feel it clutters your repo :), although I still feel something should be done but I have no clue how...

[nlharris]

Should we keep this open or close it?

[glyg]

Feel free to close it, I'll come back here if and when I get to work on it

[elichad]

Happened to open this issue by chance - I'm currently working on using LinkML to represent a validation schema for RO-Crates (via converting it to SHACL and plugging the SHACL into an RO-Crate validation tool that was developed recently). This is perhaps a slightly different use case than using LinkML schema to generate RO-Crates as @glyg is looking for, but still worth mentioning.

To be more specific, I'm currently trying to get a LinkML version of the Workflow RO-Crate profile working, because that's a lot smaller than the whole base spec to try and get a proof of concept going.

[dalito]

The description does not point out any missing feature or observed issue. So it would better be a discussion than an issue.
Instead of closing, we can convert the issue to a discussion.

[elichad]

Happy to convert to discussion. To keep things moving, I'd be interested in getting an answer to @ptsefton 's question here: #1385 (comment)

I'll also note that while I haven't found any issues specific to RO-Crate support, I am finding issues with the SHACL generator which are affecting my ability to use LinkML to create SHACL for RO-Crate validation. See:

• SHACL generation ignores all_of, none_of, and exactly_one_of #2400
• SHACL annotations are skipped on slots/properties where any_of is used #2417

(I'll update this list if I find more)