Allow prefixes pass-thru.

This PR allows a path to a prefixes file to passed
on the command line to semsql; e.g.

    semsql make foo.db -P /path/to/my/prefixes.csv

This will be used in place of the default prefix map.
Note that it is expected that standard prefixes for owl, rdf, xsd
will be passed through.

Author: cmungall

ontologies.Makefile

@@ -21,7 +21,7 @@ db/ncit.owl: download/ncit.owl
 
 
 download/fma.owl: STAMP
-	curl -L -s https://data.bioontology.org/ontologies/FMA/submissions/29/download?apikey=8b5b7825-538d-40e0-9e9e-5ab9274a9aeb > $@.tmp
+	curl -L -s http://sig.biostr.washington.edu/share/downloads/fma/release/latest/fma.owl > $@.tmp
 	sha256sum -b $@.tmp > $@.sha256
 	mv $@.tmp $@
 
@@ -75,6 +75,17 @@ db/msio.owl: download/msio.owl
 	cp $< $@
 
 
+download/modl.owl: STAMP
+	curl -L -s https://raw.githubusercontent.com/Data-Semantics-Laboratory/modular-ontology-design-library/master/MODL.owl > $@.tmp
+	sha256sum -b $@.tmp > $@.sha256
+	mv $@.tmp $@
+
+.PRECIOUS: download/modl.owl
+
+db/modl.owl: download/modl.owl
+	cp $< $@
+
+
 download/phenio.owl: STAMP
 	curl -L -s https://github.com/monarch-initiative/phenio/releases/latest/download/phenio.owl > $@.tmp
 	sha256sum -b $@.tmp > $@.sha256
@@ -130,6 +141,17 @@ db/reacto.owl: download/reacto.owl
 	cp $< $@
 
 
+download/bcio.owl: STAMP
+	curl -L -s http://humanbehaviourchange.org/ontology/bcio.owl > $@.tmp
+	sha256sum -b $@.tmp > $@.sha256
+	mv $@.tmp $@
+
+.PRECIOUS: download/bcio.owl
+
+db/bcio.owl: download/bcio.owl
+	cp $< $@
+
+
 download/go.owl: STAMP
 	curl -L -s http://purl.obolibrary.org/obo/go/extensions/go-plus.owl > $@.tmp
 	sha256sum -b $@.tmp > $@.sha256
@@ -503,4 +525,4 @@ download/%.owl: STAMP
 db/%.owl: download/%.owl
 	robot merge -i $< -o $@
 
-EXTRA_ONTOLOGIES = chiro ncit fma maxo foodon chebiplus msio phenio comploinc bero aio reacto go go-lego bao orcid cpont biolink biopax enanomapper mlo ito reactome-Homo-sapiens efo hcao edam sweetAll lov schema-dot-org cellosaurus cosmo dbpendiaont co_324 hgnc.genegroup hgnc dictybase eccode uniprot rhea swisslipid drugbank drugcentral complexportal drugmechdb rxnorm
+EXTRA_ONTOLOGIES = chiro ncit fma maxo foodon chebiplus msio modl phenio comploinc bero aio reacto bcio go go-lego bao orcid cpont biolink biopax enanomapper mlo ito reactome-Homo-sapiens efo hcao edam sweetAll lov schema-dot-org cellosaurus cosmo dbpendiaont co_324 hgnc.genegroup hgnc dictybase eccode uniprot rhea swisslipid drugbank drugcentral complexportal drugmechdb rxnorm

src/semsql/builder/build.Makefile

@@ -33,24 +33,29 @@ help:
 # All dbs are made from an initial template containing
 # (1) prefixes
 # (2) SQL Schema (primarily views)
-$(TEMPLATE): $(THIS_DIR)/sql_schema/semsql.sql build_prefixes
-	cat $< | sqlite3 $@.tmp && \
-	echo .exit | sqlite3 -echo $@.tmp -cmd ".mode csv" -cmd ".import $(THIS_DIR)/prefixes/prefixes.csv prefix" && \
-	mv $@.tmp $@
-.PRECIOUS: $(TEMPLATE)
+#$(TEMPLATE): $(THIS_DIR)/sql_schema/semsql.sql build_prefixes
+#	cat $< | sqlite3 [email protected] && \
+#	echo .exit | sqlite3 -echo [email protected] -cmd ".mode csv" -cmd ".import $(THIS_DIR)/prefixes/prefixes.csv prefix" && \
+#	mv [email protected] $@
+#.PRECIOUS: $(TEMPLATE)
 
 %-min.owl: %.owl
 	robot \
 	  remove -i $< --axioms "equivalent disjoint annotation" \
 	  filter --exclude-terms $(THIS_DIR)/exclude-terms.txt \
 	  -o $@
 
+PREFIX_CSV_PATH = $(PREFIX_DIR)/prefixes.csv
+PREFIX_YAML_PATH = $(PREFIX_DIR)/prefixes.yaml
+
 # -- MAIN TARGET --
 # A db is constructed from
 # (1) triples loaded using rdftab
 # (2) A relation-graph TSV
-%.db: %.owl %-$(RGSUFFIX).tsv $(TEMPLATE)
-	cp $(TEMPLATE) $@.tmp && \
+%.db: %.owl %-$(RGSUFFIX).tsv $(PREFIX_CSV_PATH)
+	rm -f $@.tmp && \
+	cat $(THIS_DIR)/sql_schema/semsql.sql | sqlite3 $@.tmp && \
+	echo .exit | sqlite3 -echo $@.tmp -cmd ".mode csv" -cmd ".import $(PREFIX_CSV_PATH) prefix" && \
 	rdftab $@.tmp < $< && \
 	sqlite3 $@.tmp -cmd '.separator "\t"' ".import $*-$(RGSUFFIX).tsv entailed_edge" && \
 	gzip -f $*-$(RGSUFFIX).tsv && \
@@ -61,22 +66,14 @@ $(TEMPLATE): $(THIS_DIR)/sql_schema/semsql.sql build_prefixes
 
 # -- ENTAILED EDGES --
 # relation-graph is used to compute entailed edges.
-#
-# this currently requires a few different steps, because
-#  - RG currently outputs TTL
-#  - We need a TSV using correct prefixes/CURIEs to load into our db
-#
-# will be simplified in future. See:
-#  - https://github.com/balhoff/relation-graph/issues/123
-#  - https://github.com/balhoff/relation-graph/issues/25
-%-$(RGSUFFIX).tsv: %-min.owl %-properties.txt $(PREFIX_DIR)/prefixes.yaml
+%-$(RGSUFFIX).tsv: %-min.owl %-properties.txt $(PREFIX_YAML_PATH)
 	$(RG) --disable-owl-nothing true \
                        --ontology-file $<\
 		       $(RG_PROPERTIES) \
                        --output-file [email protected] \
                        --equivalence-as-subclass true \
                        --mode TSV \
-		       --prefixes $(PREFIX_DIR)/prefixes.yaml \
+		       --prefixes $(PREFIX_YAML_PATH) \
 	               --output-individuals true \
 	               --output-subclasses true \
                        --reflexive-subclasses true && \
@@ -105,6 +102,6 @@ $(PREFIX_DIR)/prefixes.csv: $(PREFIX_DIR)/prefixes_curated.csv $(PREFIX_DIR)/pre
 	cat $^ > $@
 
 # see https://github.com/INCATools/relation-graph/issues/168
-$(PREFIX_DIR)/prefixes.yaml: $(PREFIX_DIR)/prefixes.csv
+$(PREFIX_YAML_PATH): $(PREFIX_CSV_PATH)
 	grep -v ^prefix, $< | grep -v ^obo, | perl -npe 's@,(.*)@: "$$1"@'  > $@.tmp && mv $@.tmp $@
 

src/semsql/builder/builder.py

@@ -28,12 +28,13 @@ class DockerConfig:
     memory: str = None
 
 
-def make(target: str, docker_config: Optional[DockerConfig] = None):
+def make(target: str, docker_config: Optional[DockerConfig] = None, prefix_csv_path = None):
     """
     Builds a target such as a SQLite file using the build.Makefile
 
     :param target: Make target
     :param docker_config: if passed, use ODK docker with the specific config
+    :param prefix_csv_path:
     """
     path_to_makefile = str(this_path / "build.Makefile")
     if docker_config is not None:
@@ -60,6 +61,8 @@ def make(target: str, docker_config: Optional[DockerConfig] = None):
     else:
         pre = []
     cmd = pre + ["make", target, "-f", path_to_makefile]
+    if prefix_csv_path:
+        cmd += [f"PREFIX_CSV_PATH={prefix_csv_path}"]
     logging.info(f"CMD={cmd}")
     subprocess.run(cmd)
 

src/semsql/builder/cli.py

@@ -33,7 +33,12 @@ def main(verbose: int, quiet: bool):
     show_default=True,
     help="Uses ODK docker image",
 )
-def make(path, docker):
+@click.option(
+    "--prefix-csv-path",
+    "-P",
+    help="path to csv of prefix expansions"
+)
+def make(path, docker, **kwargs):
     """
     Makes a specified target, such as a db file
 
@@ -47,7 +52,7 @@ def make(path, docker):
         docker_config = builder.DockerConfig()
     else:
         docker_config = None
-    builder.make(path, docker_config=docker_config)
+    builder.make(path, docker_config=docker_config, **kwargs)
 
 
 @main.command()

src/semsql/builder/prefixes/prefixes.csv

@@ -54,8 +54,10 @@ prefix,base
 FMA,http://purl.org/sig/ont/fma/fma
 MSIO,http://purl.obolibrary.org/obo/MSIO_
 nmrCV,http://nmrML.org/nmrCV#NMR:
+modl,https://archive.org/services/purl/purl/modular_ontology_design_library#
 biolink,https://w3id.org/biolink/vocab/
 loinc,https://loinc.org/
+BCIO,http://humanbehaviourchange.org/ontology/BCIO_
 orcid,https://orcid.org/
 evs.ncit,http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#
 old.fix,http://purl.org/obo/owl/FIX#

src/semsql/builder/prefixes/prefixes_local.csv

@@ -2,8 +2,10 @@ prefix,base
 FMA,http://purl.org/sig/ont/fma/fma
 MSIO,http://purl.obolibrary.org/obo/MSIO_
 nmrCV,http://nmrML.org/nmrCV#NMR:
+modl,https://archive.org/services/purl/purl/modular_ontology_design_library#
 biolink,https://w3id.org/biolink/vocab/
 loinc,https://loinc.org/
+BCIO,http://humanbehaviourchange.org/ontology/BCIO_
 orcid,https://orcid.org/
 evs.ncit,http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#
 old.fix,http://purl.org/obo/owl/FIX#

src/semsql/builder/registry/ontologies.yaml

@@ -9,8 +9,7 @@ ontologies:
     url: http://purl.obolibrary.org/obo/ncit.owl
     build_command: "robot relax -i $< merge -o $@"
   fma:
-    # note: this is the public API key on the main bioportal site
-    url: "https://data.bioontology.org/ontologies/FMA/submissions/29/download?apikey=8b5b7825-538d-40e0-9e9e-5ab9274a9aeb"
+    url: http://sig.biostr.washington.edu/share/downloads/fma/release/latest/fma.owl
     prefixmap:
       FMA: http://purl.org/sig/ont/fma/fma
   maxo:
@@ -28,6 +27,10 @@ ontologies:
     prefixmap:
       MSIO: http://purl.obolibrary.org/obo/MSIO_
       nmrCV: "http://nmrML.org/nmrCV#NMR:"
+  modl:
+    url: https://raw.githubusercontent.com/Data-Semantics-Laboratory/modular-ontology-design-library/master/MODL.owl
+    prefixmap:
+      modl: "https://archive.org/services/purl/purl/modular_ontology_design_library#"
   phenio:
     description: Monarch Phenomics Integrated Ontology
     url: https://github.com/monarch-initiative/phenio/releases/latest/download/phenio.owl
@@ -51,6 +54,10 @@ ontologies:
     url: https://raw.githubusercontent.com/berkeleybop/artificial-intelligence-ontology/main/aio.owl
   reacto:
     url: http://purl.obolibrary.org/obo/go/extensions/reacto.owl
+  bcio:
+    url: http://humanbehaviourchange.org/ontology/bcio.owl
+    prefixmap:
+      BCIO: http://humanbehaviourchange.org/ontology/BCIO_
   go:
     url: http://purl.obolibrary.org/obo/go/extensions/go-plus.owl
   go-lego:

tests/integration/Makefile

@@ -0,0 +1,20 @@
+RUN = poetry run
+
+input/%.owl:
+	robot merge -I http://purl.obolibrary.org/obo/$*.owl -o $@
+.PRECIOUS: input/%.owl
+
+fake/%.owl: input/%.owl
+	cp $< $@
+
+input/%.db: input/%.owl
+	$(RUN) semsql make $@
+
+fake/%.db: fake/%.owl
+	$(RUN) semsql make $@ -P conf/bad-prefixes.csv
+
+test-bad-prefix: fake/hsapdv.db
+	runoak -i $< info i^HsapDvFAKE: | head -5 | grep HsapDvFAKE && echo pass
+
+input/%.tree: input/%.db
+	runoak -i $< tree -p i,p .all -o $@

tests/integration/README.md

@@ -0,0 +1,5 @@
+# integration tests
+
+These are not currently executed as part of actions, they require installation of RG and rdftab.
+
+Note that these will be folded into the main unit tests after: https://github.com/INCATools/semantic-sql/issues/41