update HACCP_term regex to required FOODON, add multivalue example

[mslarae13]

Address syntax match to examples
Update regexs for MIxS

Based on the description for HACCP this requires the FOODON ontology.
description: Hazard Analysis Critical Control Points (HACCP) food safety terms;
This field accepts terms listed under HACCP guide food safety term (http://purl.obolibrary.org/obo/FOODON_03530221)

While this doesn't perform any validation to check if what's been entered is really in FOODON, it does some string check.

[mslarae13]

This regex requires FOODON ontology. Is this what we want?

[mslarae13]

@turbomam @only1chunts

[turbomam]

Thanks! Please use the pattern ^(\S[^\r\n]*) [FOODON:\d{7,8}]$ instead of ^.+\s*\[FOODON:\d+\](;\s*\[FOODON:\d+\])*$ or see my notes on dynamic enumerations.

[turbomam]

Did you intentionally remove the white-space between the label and the term id? I don't think that's consistent with other ontology term patterns in MIxS

[mslarae13]

@turbomam
For the white space, do you mean if it should be "lead poisoning [FOODON:03530243]" vs "lead poisoning[FOODON:03530243]"

So, the white space is supposed to be there? I thought I had it set to be valid with or without it... does it matter? If so, I'll make sure I correct it. Just tell me which is correct.

Looking at the submission schema the white space should be there. So I can make that update to the regex.

[mslarae13]

From your comment here : #802 (comment)

^(\S[^\r\n]*) [FOODON:\d{7,8}]$

I f we want to use pattern-only validation, I suggest we go with that.

That regex ^(\S[^\r\n]*) [FOODON:\d{7,8}]$
is showing me that "lead poisoning [FOODON:03530243]" is invalid... :(

... are you sure that's right?? Or am I missing something about the formatting of the value for "lead poisoning [FOODON:03530243]" ?

I think it needs to be ^.+\s*\[FOODON:\d{7,8}]$

(https://www.ebi.ac.uk/ols4/ontologies/foodon/classes/http%253A%252F%252Fpurl.obolibrary.org%252Fobo%252FFOODON_03530221)

[mslarae13]

[mslarae13]

I didn't include an example. I am not at all familiar with the FoodAnimalAndAnimalFeed extension. Before I committed time to getting familair and making an example, I wanted to check that this was a good change.

[turbomam]

Thanks @mslarae13. This is good progress. We can refine it a little:

First of all, how long are the numeric portions of FOODON URIs?

I used ChatGPT 4 to help me with that SAPRQL query

7 or 8, after subtracting the 38 characters in the base portion or the URIs, "http://purl.obolibrary.org/obo/FOODON_"

Next I asked ChatGPT 4

I want to write a regular expression for a FOODON label followed by one white-space and then a FOODON CURIe. The CURIes should be enclosed in square brackets. They start with "FOODON:" and are followed by 7 or 8 digits. The label must start with a non-white-space character but can have any number of any characters after that, as long as they aren't carriage returns, line feeds, etc.

after a little testing with regexr, we came up with

^(\S[^\r\n]*) [FOODON:\d{7,8}]$

I f we want to use pattern-only validation, I suggest we go with that.

[turbomam]

That doesn't check that the label and id portion match, etc., and it doesn't limit the choices to sub-classes of haccp guide food safety term

A better LinkML validation strategy for this might be a dynamic enumeration. They are expressed with logic, but can be expanded to an enumeration with explicit permissible values. A limitation right now is that be that the permissible values won't include the label and the id won't be enclosed in square brackets. But I would like to use this case to motivate improvements to LinkML dynamic enumerations in support of MIxS.

[turbomam]

The vskit command from the Ontology Access Kit can be used like this

vskit expand -s schema.yaml -o schema_expanded.yaml

to expand this

enums:
  HaccpTerm:
    reachable_from:
      source_ontology: bioregistry:foodon
      source_nodes:
      - FOODON:03530221   ## haccp guide food safety term
      is_direct: false
      relationship_types:
      - rdfs:subClassOf

into this

enums:
  HaccpTerm:
    reachable_from:
      source_ontology: bioregistry:foodon
      source_nodes:
      - FOODON:03530221   ## haccp guide food safety term
      is_direct: false
      relationship_types:
      - rdfs:subClassOf
    permissible_values:
      FOODON:03530231:
        text: FOODON:03530231
        meaning: FOODON:03530231
        title: hazard 3
      FOODON:03530244:
        text: FOODON:03530244
        meaning: FOODON:03530244
        title: sodium tripolyphosphate
      FOODON:03530237:
        text: FOODON:03530237
        meaning: FOODON:03530237
        title: hazard 9

[turbomam]

If using this mechanism sounds promising to you, and you want the OAK code to be modified to emit "sodium tripolyphosphate [FOODON:03530244]" instead of "FOODON:03530244", please up-vote this

• expand value set command should allow for configurable materialized values INCATools/ontology-access-kit#622

[turbomam]

Thanks! Please use the pattern ^(\S[^\r\n]*) [FOODON:\d{7,8}]$ instead of ^.+\s*\[FOODON:\d+\](;\s*\[FOODON:\d+\])*$ or see my notes on dynamic enumerations.

[only1chunts]

I agree that the change is suitable. As for the actual patturn being used, I bow to @turbomam's greater expertise on that!
The additional idea of using some sort of automated expansion thingy sounds like a good idea to me, so I have thumbs-up'd that ticket in the ontology access toolkit repo.

[only1chunts]

the changes seem reasonable to me, and I trust the combination of @turbomam and @mslarae13 to get the patturns correct (I dont have the expertise to know whats right).