Enforce taxon constraints

We add a new test to the test suite to check that taxon constraints are
not violated.

The check is done by: (1) merging the edit file with an ontology
containing disjointness axioms over sibling taxa
(obo:ncbitaxon/subsets/taxslim-disjoint-over-in-taxon.owl), (2)
expanding any macro in the merged ontology (needed for taxon constraints
depending on RO:0002161 and RO:0002175), and (3) reasoning over the
result.

This necessitates two updates to CL's ODK configuration: adding a mirror
for the taxslim-disjoint-over-in-taxon.owl module, and making sure that
the OMO:0002000 annotation property (which defines the expansion of
RO:0002161 and RO:0002175)  is not stripped away by the imports
pipeline.

Because the TC check in Uberon had been broken between March and August
2024, a TC violation has crept in that can now be found in the Uberon
import module. That violation has already been fixed upstream in Uberon,
so it will automatically disappear the next time we refresh the imports
after the next Uberon release.

In the meantime, we hack the merged_imports module to forcefully remove
the offending axiom.

.github/workflows/qc.yml

@@ -19,15 +19,45 @@ jobs:
   ontology_qc:
     # The type of runner that the job will run on
     runs-on: ubuntu-latest
-    container: obolibrary/odkfull:v1.5
+    container: obolibrary/odkfull:v1.5.2
 
     # Steps represent a sequence of tasks that will be executed as part of the job
     steps:
       # Checks-out your repository under $GITHUB_WORKSPACE, so your job can access it
       - uses: actions/checkout@v3
 
       - name: Run ontology QC checks
+        id: check
+        continue-on-error: true
         env:
           DEFAULT_BRANCH: master
         run: cd src/ontology && make ROBOT_ENV='ROBOT_JAVA_ARGS=-Xmx6G' test IMP=false PAT=false MIR=false
 
+      - name: Reason over taxon constraints
+        id: explaintc
+        continue-on-error: true
+        if: steps.check.outcome == 'failure'
+        run: |
+          if [ -s src/ontology/reports/taxon-constraint-check.txt ]; then
+            robot explain -i src/ontology/tmp/cl-plus-taxon-disjoints.ofn -M unsatisfiability -u all -r ELK -e taxon-unsats.md
+            echo "<details>\n<summary>This PR violates some taxon constraints. Here is what the reasoner has to say:</summary>\n" > comments.md
+            cat taxon-unsats.md >> comment.md
+            echo "</details>" >> comment.md
+            exit 1
+          fi
+
+      - name: Post explanation for taxon constraint violations
+        if: steps.explaintc.outcome == 'failure'
+        uses: NejcZdovc/comment-pr@v2
+        with:
+          github_token: ${{secrets.GITHUB_TOKEN}}
+          file: "../../comment.md"
+          identifier: "TAXON_CONSTRAINTS_REASONING"
+
+      # If the check failed, then the entire job should fail, even if we
+      # allowed it to continue so that we could post the message with
+      # the explanations.
+      - name: Fail on error
+        if: steps.check.outcome == 'failure'
+        run: exit 1
+

src/ontology/Makefile

@@ -1,7 +1,7 @@
 # ----------------------------------------
 # Makefile for cl
 # Generated using ontology-development-kit
-# ODK Version: v1.5
+# ODK Version: v1.5.2
 # ----------------------------------------
 # IMPORTANT: DO NOT EDIT THIS FILE. To override default make goals, use cl.Makefile instead
 
@@ -10,7 +10,7 @@
 # More information: https://github.com/INCATools/ontology-development-kit/
 
 # Fingerprint of the configuration file when this Makefile was last generated
-CONFIG_HASH=                b786b0d7cbd09184896d55b42fe68a40981419e0c9d848963a74348b7bb955b7
+CONFIG_HASH=                b30da9471aba970de63ae9153976cf8ce20f44a23d2dda7923dc04560cd0ae11
 
 
 # ----------------------------------------
@@ -46,7 +46,7 @@ REPORT_PROFILE_OPTS =
 OBO_FORMAT_OPTIONS =        
 SPARQL_VALIDATION_CHECKS =  equivalent-classes owldef-self-reference nolabels pmid-not-dbxref obsolete-replaced_by obsolete-alt-id orcid-contributor illegal-annotation-property label-synonym-polysemy illegal-date 
 SPARQL_EXPORTS =            cl_terms cl-edges cl-synonyms cl-xrefs cl-def-xrefs 
-ODK_VERSION_MAKEFILE =      v1.5
+ODK_VERSION_MAKEFILE =      v1.5.2
 
 TODAY ?=                    $(shell date +%Y-%m-%d)
 OBODATE ?=                  $(shell date +'%d:%m:%Y %H:%M')
@@ -173,7 +173,7 @@ all_main: $(MAIN_FILES)
 # ----------------------------------------
 
 
-IMPORTS =  pr go uberon ro pato ncbitaxon omo
+IMPORTS =  pr go uberon ro pato ncbitaxon ncbitaxondisjoints omo
 
 IMPORT_ROOTS =  $(IMPORTDIR)/merged_import
 IMPORT_OWL_FILES = $(foreach n,$(IMPORT_ROOTS), $(n).owl)
@@ -365,7 +365,7 @@ ALLSEED = $(PRESEED)  $(TMPDIR)/all_pattern_terms.txt \
 $(IMPORTSEED): $(ALLSEED) | $(TMPDIR)
 	if [ $(IMP) = true ]; then cat $(ALLSEED) | sort | uniq > $@; fi
 
-ANNOTATION_PROPERTIES=rdfs:label IAO:0000115 oboInOwl:is_metadata_tag 
+ANNOTATION_PROPERTIES=rdfs:label IAO:0000115 oboInOwl:is_metadata_tag OMO:0002000 
 
 # ----------------------------------------
 # Import modules
@@ -572,6 +572,13 @@ mirror-ncbitaxon: | $(TMPDIR)
 	$(ROBOT) convert -I http://purl.obolibrary.org/obo/ncbitaxon/subsets/taxslim.owl -o $(TMPDIR)/$@.owl
 
 
+## ONTOLOGY: ncbitaxondisjoints
+.PHONY: mirror-ncbitaxondisjoints
+.PRECIOUS: $(MIRRORDIR)/ncbitaxondisjoints.owl
+mirror-ncbitaxondisjoints: | $(TMPDIR)
+	$(ROBOT) convert -I http://purl.obolibrary.org/obo/ncbitaxon/subsets/taxslim-disjoint-over-in-taxon.owl -o $(TMPDIR)/$@.owl
+
+
 ## ONTOLOGY: omo
 .PHONY: mirror-omo
 .PRECIOUS: $(MIRRORDIR)/omo.owl

src/ontology/cl-odk.yaml

@@ -55,6 +55,7 @@ import_group:
     - rdfs:label
     - IAO:0000115
     - oboInOwl:is_metadata_tag
+    - OMO:0002000
   exclude_iri_patterns:
     - <http://www.informatics.jax.org/marker/MGI:*>
     - <http://purl.obolibrary.org/obo/OBA_*>
@@ -87,6 +88,8 @@ import_group:
       use_base: TRUE
     - id: ncbitaxon
       mirror_from: http://purl.obolibrary.org/obo/ncbitaxon/subsets/taxslim.owl
+    - id: ncbitaxondisjoints
+      mirror_from: http://purl.obolibrary.org/obo/ncbitaxon/subsets/taxslim-disjoint-over-in-taxon.owl
     - id: omo
       module_type: mirror
 pattern_pipelines_group:

src/ontology/cl.Makefile

@@ -170,6 +170,37 @@ $(COMPONENTSDIR)/mappings.owl: $(SRC) $(EXTERNAL_SSSOM_SETS) | all_robot_plugins
 			      --bridge-iri http://purl.obolibrary.org/obo/cl/components/mappings.owl
 
 
+# ----------------------------------------
+# TAXON CONSTRAINTS CHECK
+# ----------------------------------------
+
+# The TC check requires te taxslim-disjoint-over-in-taxon.owl ontology,
+# which is declared as an ODK mirror (under the name ncbitaxondisjoints)
+# and as such should already have been downloaded into the mirror
+# directory by the standard ODK-generated rule. But the QC checks are
+# typically run under MIR=false, so while the standard rule would be
+# invoked by Make it would not result in the mirror being downloaded.
+# So instead, we make the TC check depend on another file, and here we
+# 1. force the download of the mirror if it is not already there;
+# 2. create a link to that mirror.
+$(TMPDIR)/taxslim-disjoint-over-in-taxon.owl:
+	if [ ! -f $(MIRRORDIR)/ncbitaxondisjoints.owl ]; then \
+		$(MAKE) $(MIRRORDIR)/ncbitaxondisjoints.owl MIR=true IMP=true ; \
+	fi && \
+	ln -f -s ../$(MIRRORDIR)/ncbitaxondisjoints.owl $@
+
+# Actual TC check: we add the taxon disjointness axioms into the
+# ontology, make sure to expand all macros, then try to reason over the
+# resulting product.
+$(REPORTDIR)/taxon-constraint-check.txt: $(EDIT_PREPROCESSED) $(TMPDIR)/taxslim-disjoint-over-in-taxon.owl
+	$(ROBOT) merge $(foreach src,$^,-i $(src)) \
+		 expand -o $(TMPDIR)/cl-plus-taxon-disjoints.ofn \
+		 reason -r ELK > $@
+
+# Include the TC check in the routine tests
+test: $(REPORTDIR)/taxon-constraint-check.txt
+
+
 ##############################################
 ##### CL Template pipeline ###################
 ##############################################
@@ -321,9 +352,16 @@ $(TEMPLATEDIR)/cellxgene_subset.tsv:
 	wget $(CELLXGENE_SUBSET_URL) -O $@
 
 # Make CL-plus (CL + PCL product)
+# HACK: We have to remove the disjointness axioms between taxa, because
+# PCL does not enforce taxon constraints and may contain TC violations
+# that would be revealed here due to the presence of those axioms
+# (thereby causing that step to fail). The proper fix would be for PCL
+# to enforce TC.
+# See <https://github.com/obophenotype/provisional_cell_ontology/issues/59>
 
 cl-plus.owl: $(ONT)-full.owl
 	$(ROBOT) merge --input-iri http://purl.obolibrary.org/obo/pcl/pcl-base.owl --input $(ONT)-full.owl \
+		unmerge --input-iri http://purl.obolibrary.org/obo/ncbitaxon/subsets/taxslim-disjoint-over-in-taxon.owl \
 		reason --reasoner ELK --equivalent-classes-allowed asserted-only --exclude-tautologies structural \
 		relax \
 		reduce -r ELK \

src/ontology/imports/merged_import.owl

@@ -198565,7 +198565,6 @@ SubClassOf(<http://purl.obolibrary.org/obo/UBERON_0006293> <http://purl.obolibra
 SubClassOf(<http://purl.obolibrary.org/obo/UBERON_0006293> <http://purl.obolibrary.org/obo/UBERON_0003104>)
 SubClassOf(<http://purl.obolibrary.org/obo/UBERON_0006293> <http://purl.obolibrary.org/obo/UBERON_0004120>)
 SubClassOf(<http://purl.obolibrary.org/obo/UBERON_0006293> <http://purl.obolibrary.org/obo/UBERON_0005291>)
-SubClassOf(Annotation(<http://www.geneontology.org/formats/oboInOwl#source> "EHDAA2-modified") <http://purl.obolibrary.org/obo/UBERON_0006293> ObjectSomeValuesFrom(<http://purl.obolibrary.org/obo/BFO_0000050> <http://purl.obolibrary.org/obo/UBERON_0005602>))
 SubClassOf(Annotation(<http://www.geneontology.org/formats/oboInOwl#source> "XAO") <http://purl.obolibrary.org/obo/UBERON_0006293> ObjectSomeValuesFrom(<http://purl.obolibrary.org/obo/RO_0002202> <http://purl.obolibrary.org/obo/UBERON_0004872>))
 
 # Class: <http://purl.obolibrary.org/obo/UBERON_0006304> (future trigeminal ganglion)

src/ontology/imports/ncbitaxondisjoints_terms.txt

@@ -0,0 +1 @@
+