Submission Portal/ DH: samp_name should be required to be unique

[mslarae13]

Issue:
samp_name (1st column, sample name) validates when the values are not unique. Like the source_mat_id (globally unique id) these should be required to be unique within a submission.

Completion:
Validation rule for samp_name is changed to require it be unique

• Provide Mark with some valid and in-valid examples - Montana
• Implement validation rule changes into the submission portal - Mark
• Present updates to the team - Montana

[mslarae13]

@turbomam
Is this a schema, sheets and friends, or server issue?

[mslarae13]

@ssarrafan & @turbomam can we put this in for sprint 2? Jan 3-13? as part of the submission portal squad deliverable "Align validation with expected values"

[turbomam]

I agree that samp_name should be unique, but don't have a direct answer to your schema/sheets and friends/server issue yet. We are currently enforcing uniqueness by applying a identifier: true assertion on slot source_mat_id. Unfortunately only one slot in a class can be the identifier.

@cmungall has encouraged me to use unique_keys to express multi-field uniqueness constraints but I have dragged my feet until now.

I'm not convinced that unique_keys will solve this problem from a pure LinkML perspective, and I'm even less confident that DataHarmonizer would pick up on the unique keys.

This is currently illegal:

source_mat_id	samp_name
1	a
1	b

If we set source_mat_id and samp_name to be unique_keys, I think that will actually become legal, because there's two different composite keys.

What we really want to require is both slots are required to be individually unique:

source_mat_id	samp_name
1	a
2	b

I will have to investigate some more.

Q from @sujaypatil96: could we make the parent slot from a slot group the identifier or the unique key?

[ssarrafan]

@ssarrafan & @turbomam can we put this in for sprint 2? Jan 3-13? as part of the submission portal squad deliverable "Align validation with expected values"

Sure. I will put it under to do. @mslarae13 who should this be assigned to?

[mslarae13]

@ssarrafan Assign to me for now. I've added a couple sub tasks to get this complete. Then Mark for actual implementation.

[mslarae13]

Difficult to accomplish. LinkML only allows for 1 identifier column, and that's the source_mat_id column and also must be unique.
LinkML has unique keys to use, but they aren't checked.
Longer lived issue, backlog.

@ssarrafan with that, I recommend taking this out of the SubPort squad & this sprint, putting it into backlog. Requires LinkML updates

@turbomam , @cmungall do you agree?

[ssarrafan]

Removing from sprint, adding backlog label
@mslarae13

[turbomam]

samp_name is now considered a unique key. @pkalita-lbl has updated DataHarmonizer to apply the uniqueness constraint, but that has not appeared in a DH release yet.

[mslarae13]

The GUID (source_mat?) column should no longer be required. It's optional.. add a GUID IF you have a registered GUID. If you do not.. you'll get an NMDC minted ID.
So, we should be able to make sample_name the unique field

@pkalita-lbl will this interrupt the multiple tabs? That relies on GUID right now, right?

[pkalita-lbl]

will this interrupt the multiple tabs? That relies on GUID right now, right?

Yes, correct. We'll need to take that into account if we make that change.

[mslarae13]

When does the NMDC identifier get minted for biosamples?

• source_mat_id should be optional
• rename globally unique ID back to source material id
• multi-tab code point to samp_name

[mslarae13]

nmdc biosample identifiers minting task is separate.. captured in an issue?
submission portal to mongo connection.

[pkalita-lbl]

Schema changes are in. Once we release a new version of submission-schema we'll need to bring it into nmdc-server and make the required code changes over there.

[ssarrafan]

Moving to next sprint per @pkalita-lbl

[mslarae13]

@pkalita-lbl should this be moved to in review? Or do we need another issue for the next step?

[pkalita-lbl]

I'd rather not make a new issue since there's a lot of good context here. Paused or Blocked pending a new submission-schema release is probably a more accurate status than In Review.

[pkalita-lbl]

This is still paused waiting for a new submission-schema release.

[pkalita-lbl]

These changes are now available on data-dev.microbiomedata.org. PR: microbiomedata/nmdc-server#977