Adding usage report

Fixes #270
OBOFoundry · Aug 17, 2024 · 2970a03 · 2970a03
1 parent 47eb456
commit 2970a03
Show file tree

Hide file tree

Showing 5 changed files with 316 additions and 0 deletions.
diff --git a/src/ontology/cob.Makefile b/src/ontology/cob.Makefile
@@ -103,6 +103,25 @@ $(COB_ANNOTATIONS): $(TMPDIR)/cob-annotations.ttl
 	--annotation owl:versionInfo $(TODAY) \
 	--output $@
 
+########################################
+# -- USAGES --
+########################################
+
+# USAGE_SOURCES = ontobee ubergraph bioportal
+USAGE_SOURCES = ontobee ubergraph
+
+all_usages: $(patsubst %, reports/summary-cob-usages-%.tsv, $(USAGE_SOURCES))
+
+reports/cob-direct-subclass-counts-ontobee.tsv:
+	python3 ../scripts/generate_usage_sparql.py > $@
+
+reports/cob-usages-%.tsv: 
+	runoak -i $*: usages .idfile cob.tsv -o $@
+.PRECIOUS: reports/cob-usages-%.tsv
+
+reports/summary-cob-usages-%.tsv: reports/cob-usages-%.tsv
+	awk '{count[$$1]++} END {for (key in count) print key "\t" count[key]}' $< | sort -k2 -nr > $@
+
 
 ########################################
 # -- TESTING --

diff --git a/src/ontology/reports/cob-direct-subclass-counts-ontobee.tsv b/src/ontology/reports/cob-direct-subclass-counts-ontobee.tsv
@@ -0,0 +1,75 @@
+Class	Label	Subclass Count
+http://purl.obolibrary.org/obo/PR_000000001	protein	36130
+http://purl.obolibrary.org/obo/OBI_0000047	processed material entity	3935
+http://purl.obolibrary.org/obo/GO_0005634	cell nucleus	2782
+http://purl.obolibrary.org/obo/BFO_0000015	process	2721
+http://purl.obolibrary.org/obo/BFO_0000040	material entity	670
+http://purl.obolibrary.org/obo/GO_0032991	protein-containing macromolecular complex	540
+http://purl.obolibrary.org/obo/IAO_0000030	information	472
+http://purl.obolibrary.org/obo/IAO_0000027	data item	433
+http://purl.obolibrary.org/obo/BFO_0000023	role	395
+http://purl.obolibrary.org/obo/OBI_0000011	completely executed planned process	347
+http://purl.obolibrary.org/obo/GO_0008150	biological process	293
+http://purl.obolibrary.org/obo/IAO_0000310	document	211
+http://purl.obolibrary.org/obo/OBI_0200000	data transformation	210
+http://purl.obolibrary.org/obo/OBI_0000070	assay	209
+http://purl.obolibrary.org/obo/OBI_0100026	organism	172
+http://purl.obolibrary.org/obo/CL_0000000	cell	167
+http://purl.obolibrary.org/obo/BFO_0000016	disposition	156
+http://purl.obolibrary.org/obo/OBI_0000094	material processing	133
+http://purl.obolibrary.org/obo/ENVO_01000813	geophysical entity	91
+http://purl.obolibrary.org/obo/OBI_0000245	organization	87
+http://purl.obolibrary.org/obo/GO_0003674	gene product or complex activity	85
+http://purl.obolibrary.org/obo/BFO_0000034	function	83
+http://purl.obolibrary.org/obo/IAO_0000033	directive information entity	79
+http://purl.obolibrary.org/obo/IAO_0000104	plan specification	77
+http://purl.obolibrary.org/obo/CL_0000003	native cell	72
+http://purl.obolibrary.org/obo/IAO_0000005	objective specification	68
+http://purl.obolibrary.org/obo/OGMS_0000014	phenotypic finding	66
+http://purl.obolibrary.org/obo/ENVO_02500000	environmental process	61
+http://purl.obolibrary.org/obo/BFO_0000141	immaterial entity	61
+http://purl.obolibrary.org/obo/BFO_0000029	site	58
+http://purl.obolibrary.org/obo/MOP_0000543	physico-chemical process	49
+http://purl.obolibrary.org/obo/BFO_0000020	characteristic	49
+http://purl.obolibrary.org/obo/CHEBI_33250	uncharged atom	49
+http://purl.obolibrary.org/obo/OGMS_0000073	disease diagnosis	42
+http://purl.obolibrary.org/obo/ENVO_01001110	ecosystem	37
+http://purl.obolibrary.org/obo/BFO_0000017	realizable	36
+http://purl.obolibrary.org/obo/PATO_0000125	mass	33
+http://purl.obolibrary.org/obo/OBI_0000066	investigation	32
+http://purl.obolibrary.org/obo/PCO_0000000	collection of organisms	18
+http://purl.obolibrary.org/obo/OGMS_0000063	disease course	14
+http://purl.obolibrary.org/obo/CARO_0001008	gross anatomical part	13
+http://purl.obolibrary.org/obo/CHEBI_36342	subatomic particle	11
+http://purl.obolibrary.org/obo/CHEBI_33696	nucleic acid polymer	11
+http://purl.obolibrary.org/obo/COB_0000032	geographical location	10
+http://purl.obolibrary.org/obo/OBI_0001909	conclusion based on data	7
+http://purl.obolibrary.org/obo/CHEBI_33252	atomic nucleus	5
+http://purl.obolibrary.org/obo/COB_0000082	planned process	5
+http://purl.obolibrary.org/obo/CHEBI_24867	monoatomic ion	5
+http://purl.obolibrary.org/obo/COB_0000013	molecular entity	4
+http://purl.obolibrary.org/obo/COB_0000080	complex of molecular entities	4
+http://purl.obolibrary.org/obo/SEPIO_0000048	agent role	3
+http://purl.obolibrary.org/obo/PATO_0002193	charge	3
+http://purl.obolibrary.org/obo/CARO_0010004	cellular organism	3
+http://purl.obolibrary.org/obo/COB_0000011	atom	2
+http://purl.obolibrary.org/obo/COB_0000020	subcellular structure	2
+http://purl.obolibrary.org/obo/CL_0001034	cell in vitro	2
+http://purl.obolibrary.org/obo/OBI_0000260	plan	2
+http://purl.obolibrary.org/obo/CHEBI_10545	electron	1
+http://purl.obolibrary.org/obo/COB_0000120	information representation	1
+http://purl.obolibrary.org/obo/CHEBI_24636	proton	0
+http://purl.obolibrary.org/obo/CHEBI_30222	neutron	0
+http://purl.obolibrary.org/obo/COB_0000005	obsolete_elementary charge	0
+http://purl.obolibrary.org/obo/COB_0000014	obsolete macromolecular entity	0
+http://purl.obolibrary.org/obo/COB_0000056	immaterial anatomical entity	0
+http://purl.obolibrary.org/obo/COB_0000073	gene product	0
+http://purl.obolibrary.org/obo/COB_0000077	action specification	0
+http://purl.obolibrary.org/obo/COB_0000083	failed planned process	0
+http://purl.obolibrary.org/obo/COB_0000116	cellular membrane	0
+http://purl.obolibrary.org/obo/COB_0000119	evaluant role	0
+http://purl.obolibrary.org/obo/COB_0000121	measurement datum	0
+http://purl.obolibrary.org/obo/COB_0000122	physical information carrier	0
+http://purl.obolibrary.org/obo/COB_0001000	exposure of organism	0
+http://purl.obolibrary.org/obo/COB_0001300	device	0
+http://purl.obolibrary.org/obo/DRON_0000005	drug product	0
diff --git a/src/ontology/reports/summary-cob-usages-ontobee.tsv b/src/ontology/reports/summary-cob-usages-ontobee.tsv
@@ -0,0 +1,77 @@
+PR:000000001	36181
+OBI:0000047	3944
+GO:0005634	2955
+BFO:0000015	2746
+BFO:0000040	741
+CL:0000000	688
+GO:0032991	560
+IAO:0000030	507
+IAO:0000027	474
+BFO:0000023	415
+GO:0008150	411
+OBI:0000011	359
+OBI:0100026	244
+OBI:0000070	226
+IAO:0000310	224
+OBI:0200000	221
+BFO:0000016	164
+OBI:0000094	150
+OBI:0000245	111
+GO:0003674	106
+CHEBI:10545	106
+CL:0000003	99
+ENVO:01000813	96
+BFO:0000034	95
+IAO:0000104	89
+IAO:0000033	83
+IAO:0000005	77
+OGMS:0000014	72
+BFO:0000029	70
+CHEBI:33250	67
+ENVO:02500000	66
+BFO:0000020	65
+BFO:0000141	63
+PATO:0000125	52
+MOP:0000543	51
+OGMS:0000073	48
+OBI:0000066	48
+CHEBI:36342	47
+ENVO:01001110	45
+BFO:0000017	40
+CHEBI:33696	33
+PCO:0000000	27
+OGMS:0000063	23
+CARO:0001008	21
+CHEBI:24867	18
+CHEBI:24636	14
+COB:0000032	13
+CHEBI:33252	13
+OBI:0001909	12
+PATO:0002193	8
+COB:0000082	8
+COB:0000013	8
+SEPIO:0000048	7
+OBI:0000260	7
+COB:0000080	6
+"	6
+COB:0000077	5
+COB:0000011	5
+CL:0001034	5
+CARO:0010004	5
+COB:0000020	4
+COB:0001300	3
+COB:0000120	3
+COB:0000073	2
+CHEBI:30222	2
+used_id	1
+ID	1
+DRON:0000005	1
+COB:0001000	1
+COB:0000122	1
+COB:0000121	1
+COB:0000119	1
+COB:0000116	1
+COB:0000083	1
+COB:0000056	1
+COB:0000014	1
+COB:0000005	1
diff --git a/src/ontology/reports/summary-cob-usages-ubergraph.tsv b/src/ontology/reports/summary-cob-usages-ubergraph.tsv
@@ -0,0 +1,76 @@
+PR:000000001	21803
+BFO:0000017	4554
+GO:0005634	2961
+CL:0000000	930
+CARO:0010004	779
+CHEBI:10545	759
+CHEBI:33252	754
+GO:0008150	588
+GO:0032991	485
+BFO:0000040	450
+COB:0001000	141
+ENVO:01000813	140
+OBI:0000070	129
+BFO:0000034	129
+BFO:0000015	121
+OBI:0000011	84
+OBI:0000047	82
+OBI:0100026	69
+IAO:0000030	67
+ENVO:02500000	65
+OBI:0000094	60
+CHEBI:33250	58
+BFO:0000020	58
+IAO:0000027	53
+OBI:0200000	49
+GO:0003674	49
+CHEBI:33696	49
+BFO:0000023	49
+ENVO:01001110	37
+IAO:0000104	32
+IAO:0000033	27
+IAO:0000005	27
+OBI:0000066	25
+CHEBI:24867	23
+OBI:0001909	20
+OBI:0000245	20
+IAO:0000310	20
+BFO:0000016	19
+PATO:0000125	18
+CHEBI:36342	17
+COB:0000013	15
+OGMS:0000063	11
+COB:0000011	11
+PCO:0000000	10
+PATO:0002193	10
+OBI:0000260	10
+COB:0000082	10
+COB:0000120	9
+BFO:0000141	9
+BFO:0000029	9
+COB:0001300	8
+COB:0000121	8
+COB:0000080	8
+COB:0000077	8
+CL:0000003	8
+CHEBI:24636	8
+SEPIO:0000048	6
+OGMS:0000073	6
+COB:0000020	6
+CARO:0001008	6
+COB:0000056	5
+CL:0001034	5
+OGMS:0000014	4
+MOP:0000543	4
+DRON:0000005	4
+COB:0000122	4
+COB:0000032	4
+COB:0000073	3
+CHEBI:30222	3
+COB:0000119	2
+used_id	1
+ID	1
+COB:0000116	1
+COB:0000083	1
+COB:0000014	1
+COB:0000005	1
diff --git a/src/scripts/generate_usage_sparql.py b/src/scripts/generate_usage_sparql.py
@@ -0,0 +1,69 @@
+import requests
+
+labels = {}
+
+def expand(curie: str) -> str:
+    return f"http://purl.obolibrary.org/obo/{curie.replace(':', '_')}"
+
+def read_uris_from_file(filename):
+    tuples = []
+    with open(filename, 'r') as file:
+        for line in file:
+            line = line.strip()
+            if line.startswith("ID"):
+                continue
+            toks = line.split("\t")
+            curie = toks[0]
+            label = toks[1]
+            uri = expand(curie)
+            labels[uri] = label
+            tuples.append((uri, label))
+    return tuples
+
+def generate_sparql_query(uris):
+    values_clause = "\n    ".join(f"<{uri}>" for uri in uris)
+    return f"""PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
+PREFIX obo: <http://purl.obolibrary.org/obo/>
+
+SELECT ?class (COUNT(DISTINCT ?subclass) AS ?subclassCount)
+WHERE {{
+  VALUES ?class {{
+    {values_clause}
+  }}
+  ?subclass rdfs:subClassOf ?class .
+  FILTER (?subclass != ?class)
+}}
+GROUP BY ?class
+ORDER BY DESC(?subclassCount)"""
+
+def execute_sparql_query(query):
+    endpoint = "https://sparql.hegroup.org/sparql"
+    headers = {
+        "Accept": "application/sparql-results+json",
+        "Content-Type": "application/x-www-form-urlencoded"
+    }
+    data = {
+        "query": query,
+        "format": "json"
+    }
+    response = requests.post(endpoint, headers=headers, data=data)
+    return response.json()
+
+# Usage
+pairs = read_uris_from_file('cob.tsv')
+uris = [p[0] for p in pairs]
+query = generate_sparql_query(uris)
+
+results = execute_sparql_query(query)
+
+# Process and print results
+print("Class\tLabel\tSubclass Count")
+for result in results['results']['bindings']:
+    class_uri = result['class']['value']
+    # label = result['label']['value']
+    label = labels[class_uri]
+    count = result['subclassCount']['value']
+    print(f"{class_uri}\t{label}\t{count}")
+    del labels[class_uri]
+for class_uri, label in labels.items():
+    print(f"{class_uri}\t{label}\t0")