“DNA Sample ID” should be unique

[pvangay]

These values are pre-determined by JGI, but having validation for uniqueness can help in case the user accidentally overwrites the values.

[turbomam]

I guess I should have said that I had anticipated this. Presumably it will apply to RNA Sample ID as well.

DH takes its uniqueness cues from the LinkML identifier attribute on each slot that should become a DH column. Unfortunately, the DH columns represent fields/slots on one single class, and each class can only have identifier slot.

@cmungall It looks like there are some additional constructs in LinkML for communicating unique values required. I'll have to familiarize myself with them.

We could also come up with a new string serialization (SNTC pattern) for cases like this like {unique text}

@sujaypatil96 Either of those techniques will require changes to SNTC (which I consider acceptable up until days before GSP) and or changes to linkml2dataharmonizer.py, arround here (which I was hoping to be more conservative about)

[pkalita-lbl]

We can use the unique_keys construct in LinkML for this. DataHarmonizer understands that construct now in its validation code.