Fix data model pending Schemativ v23.10.1

[adamjtaylor]

As discussed in the 2023-10-10 HTAN ops meeting we will make a fix release in our data model rather than waiting for the backend Schematic v23.10.1 fix which is not scheduled until the end of October.

Once merged we will make a new release and deploy to DCA.

From FDS-1174

. If not, you could create a temp fix release that remove File and Publication from the DependsOnComponent column of the data model from lines #123 Mass Spec Aux File and #842 Publication Manifest.

[adamjtaylor]

Our line numbers do not match up with the comments here.

Mass Spectrometry Auxiliary file is currently our line 170. No entry in depends on component so looks to have been fixed in #289

Publication Manifest currently on line 882. Removed Publication from DependsOnComponent

[adamjtaylor]

Re-opening as this didn't seem to fix the issue.

Re-reading the linked ticket

The reason there is a namespace clash is bc we currently build the data model as an extension of Biothings. This is largely invisible to users (and causes unintended errors like this) and doesn't serve the function it was initially intended anymore so we are not including it in the initial release of the new schema refactor. We will allow similar functionality in the future, but users will opt in. So in the future accidental namespace clashes with a base model will not be an issue.

Additionally the way the refactor works internally, we already account if a display name isnt recorded and use the label instead, so this will be handled internally if there are different base models imported.

You can see the Biothings model at schematic/schematic/etc/data_models/biothings.model.jsonld that will guide in looking for namespace overlaps.

Metabolite is a single namespace clash there could be others, which would affect when the issue would have been noticed, based on what manifests any ‘bad’ names pulled into. We saw the issue (referenced in FDS-1030) pop up with a submission actually bc it couldnt find the display name.

The reason there is a namespace clash is bc we currently build the data model as an extension of Biothings. This is largely invisible to users (and causes unintended errors like this) and doesn't serve the function it was initially intended anymore so we are not including it in the initial release of the new schema refactor. We will allow similar functionality in the future, but users will opt in. So in the future accidental namespace clashes with a base model will not be an issue.

Additionally the way the refactor works internally, we already account if a display name isnt recorded and use the label instead, so this will be handled internally if there are different base models imported.

You can see the Biothings model at schematic/schematic/etc/data_models/biothings.model.jsonld that will guide in looking for namespace overlaps.

Metabolite is a single namespace clash there could be others, which would affect when the issue would have been noticed, based on what manifests any ‘bad’ names pulled into. We saw the issue (referenced in FDS-1030) pop up with a submission actually bc it couldnt find the display name.

So we need to try and remove these namespace clashes. Will try and figure out if we can remove metabolite first

[adamjtaylor]

Confirmed from testing with the schematic cli that removing protein and metabolite from Assay Type should allow Biospeciem to be generated

[adamjtaylor]

As far as I can tell protein and metabolite have not been used - they are there for use in Mass Spectrometry Level 1 but have not been used yet. Stanford are likely the only users impacted so they will need to delay and resubmission of this template until after Schematic 23.10.1 has been deployed and these elements returned and tested

[adamjtaylor]

Fix confirmed in staging