add slot usage on identifier column?

[cmungall]

can you give more context?

[turbomam]

I think we discussed this somewhere else, so I'll try to look that up. Maybe even another GH issue!

I would like to make slot usage assertions about columns that have been set as identifiers at the schema level. The use case is saying where slots should appear as columns in DataHarmonizer. We have been doing that with assertions but will be switching to slot groups and ranks.

Assume I

• define the slot, with the identifier assertion, on its own row in schemasheets
• define a class on it's own row
• use a third row, mentioning the class and the slot, to make slot usage assertions

LinkML doesn't like that, because then I have an id slot on class biosample (inherited from named thing) and biosample_id from the slot usage.

[cmungall]

are you trying to deliberately override the name to rename id and biosample_id?

I suspect what the issue is here is that the slot_usage causes a name mangling - can you say where this manifests?

[turbomam]

this issue overlaps with the broader LinkML questions

• why can't a class have multiple identifiers
• why isn't there a unique attribute for slots
• compound unique_keys ? seems complicated

[sierra-moxon]

we now have compound keys. closing as complete, please reopen if necessary. :)