-
Notifications
You must be signed in to change notification settings - Fork 206
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
DRAFT: New Principle 19 - for OBO Operations Review #2668
base: master
Are you sure you want to change the base?
Changes from 8 commits
947ddac
d314795
811787b
3b48962
31c1d85
71d4fa0
ce9052b
4515175
a11f207
File filter
Filter by extension
Conversations
Jump to
Diff view
Diff view
There are no files selected for viewing
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,144 @@ | ||
--- | ||
layout: principle | ||
id: fp-019-term-stability | ||
title: Stability of Term Meaning (principle 19) | ||
--- | ||
|
||
GO TO: [Recommendations/Requirements](#recommendations-and-requirements) | [Implementation](#implementation) | [Examples/Counter‑Examples](#examples) | [Criteria for Review](#criteria-for-review) | [Feedback/Discussion](#feedback-and-discussion) | ||
|
||
Summary | ||
------- | ||
|
||
The definition of a term MUST always denote the same thing(s)--known as "referent(s)"--in reality. If a proposed change to the definition would substantially change its referents, then a new term with new IRI and definition MUST instead be created. | ||
|
||
Purpose | ||
------- | ||
|
||
Users of an ontology depend on the stability of its terms and their meanings. Therefore, changes to the definition of a term should never substantially shift its meaning. Put another way, its set of referents MUST remain stable, within reason. That is, changes to a term definition should not cause that term to point to different entities than it denotes already. | ||
|
||
Recommendations and Requirements | ||
------- | ||
|
||
If changing a term definition would change its referents, then instead a new term MUST be created with a new IRI and the new definition. Minor changes to the definition for clarity, grammar, and/or proper punctuation that do not change the referents are permitted. What is considered a 'minor change' will likely need to be considered on a case-by-case basis; it is left to the ontology developers to decide. However, any feedback from users MUST be taken into account. | ||
|
||
Conditions under which a term must be deprecated according to this principle, or for which term deprecation should be considered, include: | ||
|
||
1) The old textual definition misses its intended target. This includes cases where the term refers to non-existent referents (as might happen, for example, when new research reveals that the referent does not exist in reality). | ||
1) The original term definition is considered sufficiently “damaged” (too vague, too restrictive, too misused or too misunderstood). | ||
|
||
In all cases the developers SHOULD provide guidance on how to handle deprecated terms (either by exact replacement or by considering other terms), and be mindful of the potential costs to users of the ontology who might use the existing term. As well, developers SHOULD pre-announce term obsoletions. See [Principle 13](http://obofoundry.org/principles/fp-013-notification.html) for guidance on such announcements. | ||
|
||
Implementation | ||
------- | ||
|
||
Detailed procedures for obsoleting a term are described on the OBO Academy page [Obsoleting an Existing Ontology Term](https://oboacademy.github.io/obook/howto/obsolete-term/). | ||
|
||
<i><b>To obsolete a term, the ontology developer</b></i> MUST: | ||
1) Mark the term as obsolete | ||
- OWL format: Add an "owl:deprecated" annotation property with value of "true^xsd:boolean" | ||
- OBO format: Add an "is_obsolete: true" tag | ||
2) Prepend the string "obsolete " (including the space) to the term label | ||
- NOTE: To be consistent with [Principle 12](https://obofoundry.org/principles/fp-012-naming-conventions.html) "Naming Conventions", the syntax/format MUST be precisely as given above (that is, the exact string as shown, lowercase and space included, with no other punctuation before or after). Thus, the following are disallowed: "Obsolete {label}", "obsolete_{label}", "OBSOLETE {label}" (and variations thereof). | ||
3) Remove all existing logical axioms from the term | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. Should be: all logical axioms referencing the term (including logical axioms about the term). Can also add an example: For example, if an ontology has A part-of B, B part-of C, and if B is obsoleted, then there should be no part-of edges incoming or outgoing from B. Edge axioms should be rewired, e.g. making A part-of C. There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. Is your suggestion different from the one below, in (4)? Is the lack of the word 'logical' in point (4) the problem? |
||
4) Remove or replace all usages of the term elsewhere in the ontology | ||
|
||
It is not necessary (and not advisable) to delete the textual definition. | ||
|
||
<i><b>To obsolete a term, the ontology developer</b></i> SHOULD: | ||
1) Indicate any exact term replacement: | ||
- OWL: Use the "term replaced by" annotation property from OMO ([IAO:0100001](http://purl.obolibrary.org/obo/IAO_0100001)) with the value set to the IRI of the relevant term | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. I recommend backticks over double quotes in markdown when referring to controlled terms, curies, etc There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. Hmm, I think we've not been terribly consistent over all the principles. I'll put this on our to-do list (that is, I'll make an issue). #2669 |
||
- OBO: Use the "replaced_by:" tag with the value set to the CURIE of the relevant term | ||
2) Indicate any inexact term replacements: | ||
- OWL: Use the "oboInOwl:consider" annotation property with the value set to the full IRI(s) of the relevant term(s) | ||
``` | ||
<oboInOwl:consider rdf:resource="http://purl.obolibrary.org/obo/OBI_0001544")> | ||
``` | ||
|
||
- OBO: Use the "consider:" tag with the value set to the CURIE(s) of the relevant term(s) | ||
``` | ||
consider: OBI:0001544 | ||
``` | ||
|
||
Note that some older implementations in OWL used the CURIE method as shown below, but this is not preferred. | ||
|
||
``` | ||
<oboInOwl:consider rdf:datatype="http://www.w3.org/2001/XMLSchema#string">OBI:0001544</oboInOwl:consider> | ||
``` | ||
<i><b>To obsolete a term, the ontology developer</b></i> MAY: | ||
|
||
1) Prepend the string "OBSOLETE. " (this precise string, including the space) to the term definition. NOTE: This MUST be implemented consistently. That is, if applied at all, it has to be applied to every obsoleted term definition. | ||
2) Indicate the reason(s) for obsoleting: | ||
- OWL: Use the "has obsolescence reason" annotation property from OMO ([IAO:0000231](http://purl.obolibrary.org/obo/IAO_0000231)) with the value set to the IRI of one of the individuals of the "obsolescence reason specification" term [IAO:0000225](http://purl.obolibrary.org/obo/IAO_0000225). See below for example. | ||
- OBO: Use "relationship:" with the CURIE for the annotation property (IAO:0000231) and a CURIE for the specific reason (an individual from the "obsolescence reason specification" term [IAO:0000225](http://purl.obolibrary.org/obo/IAO_0000225)). See below for example. Note that older implementations often used alternative methods (described after the examples). These methods are still valid, but are not preferred. | ||
|
||
Examples | ||
------- | ||
|
||
The Ontology for Biomedical Investigations obsolete term "cell lysate MHC competitive binding assay using radioactivity detection" (OBI:0001574) can be deprecated as follows: | ||
|
||
OWL: | ||
``` | ||
<owl:Class rdf:about="http://purl.obolibrary.org/obo/OBI_0001574"> | ||
<obo:IAO_0000111>obsolete cell lysate MHC competitive binding assay using radioactivity detection</obo:IAO_0000111> | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. There should be guidance specific to "editor preferred term". None of the ontologies I work on use this, it seems that it is often redundant? I would just remove it altogether it's not clear why an obsolete concept would need both this and a label, but as I say I don't use it... There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. Good point. Will discuss with those who do use that field. |
||
<obo:IAO_0000115>Competitive inhibition of binding assay measuring MHC ligand binding by radioactivity detection using MHC derived from a cell lysate</obo:IAO_0000115> | ||
<obo:IAO_0000231 rdf:resource="http://purl.obolibrary.org/obo/IAO_0000227"/> | ||
<obo:IAO_0100001 rdf:resource="http://purl.obolibrary.org/obo/OBI_0001544"/> | ||
<rdfs:label>obsolete cell lysate MHC competitive binding assay using radioactivity detection</rdfs:label> | ||
<owl:deprecated rdf:datatype="http://www.w3.org/2001/XMLSchema#boolean">true</owl:deprecated> | ||
</owl:Class> | ||
``` | ||
|
||
OBO: | ||
``` | ||
[Term] | ||
id: OBI:0001574 | ||
name: obsolete cell lysate MHC competitive binding assay using radioactivity detection | ||
def: "OBSOLETE. Competitive inhibition of binding assay measuring MHC ligand binding by radioactivity detection using MHC derived from a cell lysate." [] | ||
relationship: IAO:0000231 IAO:0000227 | ||
is_obsolete: true | ||
replaced_by: OBI:0001544 | ||
``` | ||
For OBO format, there are multiple alternatives: | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. I don't think this has anything directly to do with OBO format. GO, Uberon, CL, all phenotype ontologies all use the same simple data model e.g. making use of comments because this is what browsers show more prominently, because it's simpler and more consistent. Nothing to do with format. OBO Format can represent annotation assertions at exactly the same level of expressivity as OWL There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. These alternatives are listed for OBO format because they are used 'in the field' so to speak. That is, unlike ontologies in OWL format--which all seem to use the same deprecation methods--OBO formatted ontologies have been using different means. I've confirmed that either way is fine, so we say all this to acknowledge these alternatives. |
||
1) Use "property_value:" instead of "relationship:". | ||
1) Use a free text "comment:". | ||
1) Use the annotation property label (with underscores) instead of the CURIE "IAO:0000231", and the obsolescence reason label instead of the relevant CURIE. Note that the underscore version of the property label will need to be created in the ontology: | ||
``` | ||
[Typedef] | ||
id: has_obsolescence_reason | ||
name: has obsolescence reason | ||
xref: IAO:0000233 | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. wrong ID There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. Oops, thanks! Fixed. |
||
is_metadata_tag: true | ||
``` | ||
Then: | ||
``` | ||
relationship: has_obsolescence_reason IAO:0000227 ! terms merged | ||
``` | ||
Or: | ||
``` | ||
property_value: has_obsolescence_reason IAO:0000227 | ||
``` | ||
|
||
Counter-example | ||
------- | ||
|
||
The PRO term "phosphoprotein" (PR:000037070) is defined as "A protein that includes at least one phosphorylated residue." A study finds 2000 more examples than was previously known. In this case, no new term needs to be made (nor the original deleted) because (1) the term definition did not change; and (2) the referent--proteins with a phophoresidue--did not change (that is, the newly-discovered phosphoproteins are just additional examples of that referent). | ||
|
||
Criteria for review | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. presumably this is criteria for dashboard review? The earlier text lays out criteria about definitions changing in meaning. This could easily be checked by LLM using KGCL, but the dashboard doesn't support this yet There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. The heading 'Criteria for review' intends to capture all possible types of review. Here we list what we can reasonably check. If the future adds more powerful checks, those checks will be added. |
||
------- | ||
|
||
The OBO Dashboard will show: | ||
- An ERROR if any obsolete term (that is, a term with an "owl:deprecated" property or "is_obsolete: true" tag) does not also have 'obsolete ' (that exact string, lowercase and space included, with no other punctuation) prepended to the label | ||
- An ERROR if any obsolete term (as indicated by term label or definition) lacks an "owl:deprecated" property or "is_obsolete: true" tag | ||
- An ERROR if an obsolete term has any associated logical axioms (including any subClassOf/is_a declarations) | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. is "associated with" the same as "referencing"? (best to use precise terms, we generally follow the owlapi vocabulary e.g "about" "referencing") There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. Yes, that's half the intent. Revising. |
||
- A WARN if there is at least one term with 'OBSOLETE. ' prepended to the definition but not all obsolete terms are likewise prepended | ||
|
||
[This check is automatically validated.](checks/fp_019) | ||
|
||
Feedback and Discussion | ||
------- | ||
|
||
To suggest revisions or begin a discussion pertaining to this principle, please [create an issue on GitHub](https://github.com/OBOFoundry/OBOFoundry.github.io/issues/new?labels=attn%3A+Editorial+WG,principles&title=Principle+%2319+%22Stability%22of%22Term%22Meaning%22+%3CENTER+ISSUE+TITLE%3E). | ||
|
||
To suggest revisions or begin a discussion pertaining to the automated validation of this principle, please [create an issue on GitHub](). | ||
|
||
See also [this discussion](https://github.com/OBOFoundry/OBOFoundry.github.io/issues/597) by the OBO Foundry Operations Committee focusing on term deprecation and [this discussion](https://github.com/OBOFoundry/OBOFoundry.github.io/issues/964) regarding the proposal of this principle. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
See also kgcl:NodeObsoletion for a taxonomy of kinds of obsoletion