Ontology updates

docs/_changelog.d/combinatorial_synonym_generator.bugfix.rst

@@ -0,0 +1 @@
+Fix synonym generator to only check if strings exist in original synonyms. Update tests

docs/_changelog.d/krt.bugfix.rst

@@ -0,0 +1 @@
+Remove save/reset button not belonging on page 1

docs/_changelog.d/ontology_updates.feature.rst

@@ -0,0 +1 @@
+Update ontologies to later versions

kazu/conf/SynonymGeneration/default.yaml

@@ -58,37 +58,37 @@ meddra:
       include_greek: true
     - _target_: kazu.ontology_preprocessing.synonym_generation.TokenListReplacementGenerator
       token_lists_to_consider:
-        -  - anomalous
-           - unusual
-           - atypical
-           - deviant
-           - aberrant
-           - peculiar
-           - eccentric
-           - strange
-           - bizarre
-           - odd
-           - weird
-           - uncommon
-           - freakish
-           - irregular
-           - outlandish
-           - exceptional
-           - unnatural
-           - unhealthy
-           - abnormal
-        -  - typical
-           - ordinary
-           - standard
-           - regular
-           - usual
-           - natural
-           - routine
-           - average
-           - expected
-           - balanced
-           - stable
-           - healthy
+        - - anomalous
+          - unusual
+          - atypical
+          - deviant
+          - aberrant
+          - peculiar
+          - eccentric
+          - strange
+          - bizarre
+          - odd
+          - weird
+          - uncommon
+          - freakish
+          - irregular
+          - outlandish
+          - exceptional
+          - unnatural
+          - unhealthy
+          - abnormal
+        - - typical
+          - ordinary
+          - standard
+          - regular
+          - usual
+          - natural
+          - routine
+          - average
+          - expected
+          - balanced
+          - stable
+          - healthy
 
     - _target_: kazu.ontology_preprocessing.synonym_generation.VerbPhraseVariantGenerator
       spacy_model_path: ${SciSpacyPipeline.path}
@@ -107,112 +107,112 @@ meddra:
         - "{NOUN} {TARGET}"
       lemmas_to_consider:
         increase:
-            - increasing
-            - increased
-            - increases
+          - increasing
+          - increased
+          - increases
 
         expand:
-            - expanding
-            - expands
-            - expanded
+          - expanding
+          - expands
+          - expanded
 
         enhance:
-            - enhancing
-            - enhances
-            - enhanced
+          - enhancing
+          - enhances
+          - enhanced
         raise:
-            - raising
-            - raises
-            - raised
+          - raising
+          - raises
+          - raised
         boost:
-            - boosting
-            - boosts
-            - boosted
+          - boosting
+          - boosts
+          - boosted
         escalate:
-            - escalating
-            - escalates
-            - escalated
+          - escalating
+          - escalates
+          - escalated
         augment:
-            - augmenting
-            - augments
-            - augmented
+          - augmenting
+          - augments
+          - augmented
         intensify:
-            - intensifying
-            - intensifies
-            - intensifed
+          - intensifying
+          - intensifies
+          - intensifed
         amplify:
-            - amplifing
-            - amplifies
-            - amplified
+          - amplifing
+          - amplifies
+          - amplified
         strengthen:
-            - strengthening
-            - strengthens
-            - strengthened
+          - strengthening
+          - strengthens
+          - strengthened
         heighten:
-            - heightening
-            - heightens
-            - heightened
+          - heightening
+          - heightens
+          - heightened
         decrease:
-            - decreasing
-            - decreases
-            - decreased
+          - decreasing
+          - decreases
+          - decreased
         reduce:
-            - reducing
-            - reduces
-            - reduced
+          - reducing
+          - reduces
+          - reduced
         diminish:
-            - diminishing
-            - diminishes
-            - diminished
+          - diminishing
+          - diminishes
+          - diminished
         lower:
-            - lowering
-            - lowers
-            - lowered
+          - lowering
+          - lowers
+          - lowered
         minimize:
-            - minimizing
-            - minimizes
-            - minimized
+          - minimizing
+          - minimizes
+          - minimized
         minimise:
-            - minimising
-            - minimises
-            - minimised
+          - minimising
+          - minimises
+          - minimised
         curtail:
-            - curtailing
-            - curtails
-            - curtailed
+          - curtailing
+          - curtails
+          - curtailed
         shrink:
-            - shrinking
-            - shrinks
-            - shrinked
+          - shrinking
+          - shrinks
+          - shrinked
         contract:
-            - contracting
-            - contracts
-            - contracted
+          - contracting
+          - contracts
+          - contracted
         narrow:
-            - narrowing
-            - narrows
-            - narrowed
+          - narrowing
+          - narrows
+          - narrowed
         taper:
-            - tapering
-            - tapers
-            - tapered
+          - tapering
+          - tapers
+          - tapered
         constrict:
-            - constricting
-            - constricts
-            - constricted
+          - constricting
+          - constricts
+          - constricted
         dwindle:
-            - dwindling
-            - dwindles
-            - dwindled
+          - dwindling
+          - dwindles
+          - dwindled
         subside:
-            - subsiding
-            - subsides
-            - subsided
+          - subsiding
+          - subsides
+          - subsided
         weaken:
-            - weakening
-            - weakens
-            - weakened
+          - weakening
+          - weakens
+          - weakened
         decrement:
-            - decrementing
-            - decrements
-            - decremented
+          - decrementing
+          - decrements
+          - decremented

kazu/conf/ontologies/cellosaurus.yaml

@@ -8,7 +8,7 @@ parsers:
     _target_: kazu.ontology_preprocessing.parsers.CellosaurusOntologyParser
     name: CELLOSAURUS
     in_path: ${oc.env:KAZU_MODEL_PACK}/ontologies/cellosaurus.obo
-    data_origin: '49.0'
+    data_origin: "50.0"
     string_scorer: ${SapbertStringSimilarityScorer}
     entity_class: cell_line
     curations_path: ${oc.env:KAZU_MODEL_PACK}/ontologies/${.name}_curations.jsonl

kazu/conf/ontologies/cl.yaml

@@ -8,7 +8,7 @@ parsers:
     _target_: kazu.ontology_preprocessing.parsers.CLOntologyParser
     name: CL
     in_path: ${oc.env:KAZU_MODEL_PACK}/ontologies/cl.owl
-    data_origin: '2024-05-15'
+    data_origin: "2024-09-26"
     string_scorer: ${SapbertStringSimilarityScorer}
     synonym_generator: ${SynonymGeneration.default}
     entity_class: cell_type

kazu/conf/ontologies/go_bp.yaml

@@ -8,7 +8,7 @@ parsers:
     _target_: kazu.ontology_preprocessing.parsers.BiologicalProcessGeneOntologyParser
     name: BP_GENE_ONTOLOGY
     in_path: ${oc.env:KAZU_MODEL_PACK}/ontologies/go.owl
-    data_origin: '2024-06-17'
+    data_origin: "2024-09-08"
     string_scorer: ${SapbertStringSimilarityScorer}
     synonym_generator: ${SynonymGeneration.default}
     entity_class: biological_process

kazu/conf/ontologies/go_cc.yaml

@@ -8,7 +8,7 @@ parsers:
     _target_: kazu.ontology_preprocessing.parsers.CellularComponentGeneOntologyParser
     name: CC_GENE_ONTOLOGY
     in_path: ${oc.env:KAZU_MODEL_PACK}/ontologies/go.owl
-    data_origin: '2024-06-17'
+    data_origin: "2024-09-08"
     string_scorer: ${SapbertStringSimilarityScorer}
     synonym_generator: ${SynonymGeneration.default}
     entity_class: cellular_component

kazu/conf/ontologies/go_mf.yaml

@@ -8,7 +8,7 @@ parsers:
     _target_: kazu.ontology_preprocessing.parsers.MolecularFunctionGeneOntologyParser
     name: MF_GENE_ONTOLOGY
     in_path: ${oc.env:KAZU_MODEL_PACK}/ontologies/go.owl
-    data_origin: '2024-06-17'
+    data_origin: "2024-09-08"
     string_scorer: ${SapbertStringSimilarityScorer}
     synonym_generator: ${SynonymGeneration.default}
     entity_class: molecular_function

kazu/conf/ontologies/mondo.yaml

@@ -8,7 +8,7 @@ parsers:
     _target_: kazu.ontology_preprocessing.parsers.MondoOntologyParser
     name: MONDO
     in_path: ${oc.env:KAZU_MODEL_PACK}/ontologies/mondo.json
-    data_origin: '2024-06-04'
+    data_origin: "2024-10-15T13:28:52.008470"
     string_scorer: ${SapbertStringSimilarityScorer}
     synonym_generator: ${SynonymGeneration.default}
     entity_class: disease

kazu/conf/ontologies/ot_bio_proc.yaml

@@ -8,14 +8,14 @@ parsers:
     name: OPENTARGETS_BIOLOGICAL_PROC
     _target_: kazu.ontology_preprocessing.parsers.OpenTargetsDiseaseOntologyParser
     in_path: ${oc.env:KAZU_MODEL_PACK}/ontologies/opentargets/diseases
-    data_origin: '24.06'
+    data_origin: "24.09"
     string_scorer: ${SapbertStringSimilarityScorer}
     synonym_generator: ${SynonymGeneration.default}
     entity_class: biological_process
     curations_path: ${oc.env:KAZU_MODEL_PACK}/ontologies/${.name}_curations.jsonl
     global_actions: ${global_actions}
     allowed_therapeutic_areas:
-      - 'GO_0008150'
+      - "GO_0008150"
     autocurator: ${autocurator.basic}
     ontology_downloader:
       _target_: kazu.ontology_preprocessing.downloads.OpenTargetsOntologyDownloader

kazu/conf/ontologies/ot_disease.yaml

@@ -8,35 +8,35 @@ parsers:
     name: OPENTARGETS_DISEASE
     _target_: kazu.ontology_preprocessing.parsers.OpenTargetsDiseaseOntologyParser
     in_path: ${oc.env:KAZU_MODEL_PACK}/ontologies/opentargets/diseases
-    data_origin: '24.06'
+    data_origin: "24.09"
     string_scorer: ${SapbertStringSimilarityScorer}
     synonym_generator: ${SynonymGeneration.default}
     entity_class: disease
     curations_path: ${oc.env:KAZU_MODEL_PACK}/ontologies/${.name}_curations.jsonl
     global_actions: ${global_actions}
     allowed_therapeutic_areas:
-      - 'MONDO_0024458'
-      - 'OTAR_0000006'
-      - 'MONDO_0002025'
-      - 'EFO_0001379'
-      - 'EFO_0009690'
-      - 'EFO_0000540'
-      - 'EFO_0005741'
-      - 'EFO_0000319'
-      - 'EFO_0009605'
-      - 'MONDO_0045024'
-      - 'OTAR_0000018'
-      - 'OTAR_0000020'
-      - 'EFO_0010285'
-      - 'OTAR_0000010'
-      - 'EFO_0005803'
-      - 'EFO_0005932'
-      - 'OTAR_0000009'
-      - 'MONDO_0021205'
-      - 'OTAR_0000017'
-      - 'EFO_0010282'
-      - 'OTAR_0000014'
-      - 'EFO_0000618'
+      - "MONDO_0024458"
+      - "OTAR_0000006"
+      - "MONDO_0002025"
+      - "EFO_0001379"
+      - "EFO_0009690"
+      - "EFO_0000540"
+      - "EFO_0005741"
+      - "EFO_0000319"
+      - "EFO_0009605"
+      - "MONDO_0045024"
+      - "OTAR_0000018"
+      - "OTAR_0000020"
+      - "EFO_0010285"
+      - "OTAR_0000010"
+      - "EFO_0005803"
+      - "EFO_0005932"
+      - "OTAR_0000009"
+      - "MONDO_0021205"
+      - "OTAR_0000017"
+      - "EFO_0010282"
+      - "OTAR_0000014"
+      - "EFO_0000618"
     autocurator: ${autocurator.basic}
     ontology_downloader:
       _target_: kazu.ontology_preprocessing.downloads.OpenTargetsOntologyDownloader