Exclude Attribute, ProvActivity, ProvEntity from ER diagrams

erdiagram-autogen/anatomical_structure.md

@@ -176,21 +176,12 @@ ParcellationTerminology {
 }
 
 AnatomicalAnnotationSet ||--|o VersionedNamedThing : "revision_of"
-AnatomicalAnnotationSet ||--}o Attribute : "has attribute"
 AnatomicalSpace ||--|o VersionedNamedThing : "revision_of"
-AnatomicalSpace ||--}o Attribute : "has attribute"
 ImageDataset ||--|o VersionedNamedThing : "revision_of"
-ImageDataset ||--}o Attribute : "has attribute"
 ParcellationAtlas ||--|o VersionedNamedThing : "revision_of"
-ParcellationAtlas ||--}o Attribute : "has attribute"
 ParcellationColorScheme ||--|o VersionedNamedThing : "revision_of"
-ParcellationColorScheme ||--}o Attribute : "has attribute"
 ParcellationTerm ||--|o VersionedNamedThing : "revision_of"
-ParcellationTerm ||--}o Attribute : "has attribute"
 ParcellationTermSet ||--|o VersionedNamedThing : "revision_of"
-ParcellationTermSet ||--}o Attribute : "has attribute"
 ParcellationTerminology ||--|o VersionedNamedThing : "revision_of"
-ParcellationTerminology ||--}o Attribute : "has attribute"
 
 ```
-

erdiagram-autogen/bke_taxonomy.md

@@ -258,12 +258,6 @@ ObservationRow {
     stringList provided_by  
     label_typeList synonym  
     stringList type  
-}
-ProvActivity {
-
-}
-ProvEntity {
-
 }
 SpatialProportions {
     float adj  
@@ -292,80 +286,40 @@ SpatialProportions {
     uriorcurieList xref  
 }
 
-Abbreviation ||--|o ProvActivity : "was_generated_by"
-Abbreviation ||--|o ProvEntity : "was_derived_from"
-Abbreviation ||--}o Attribute : "has attribute"
 Abbreviation ||--}o CellTypeTaxon : "denotes_cell_type"
 Abbreviation ||--}o GeneAnnotation : "denotes_gene_annotation"
 Abbreviation ||--}o ParcellationTerm : "denotes_parcellation_term"
-CellSpecimen ||--|o ProvActivity : "was_generated_by"
-CellSpecimen ||--|o ProvEntity : "was_derived_from"
-CellSpecimen ||--}o Attribute : "has attribute"
 CellTypeSet ||--|o CellTypeSet : "has_parent"
 CellTypeSet ||--|o CellTypeTaxonomy : "part_of_taxonomy"
-CellTypeSet ||--|o ProvActivity : "was_generated_by"
-CellTypeSet ||--|o ProvEntity : "was_derived_from"
 CellTypeSet ||--}o Abbreviation : "has_abbreviation"
-CellTypeSet ||--}o Attribute : "has attribute"
 CellTypeSet ||--}o CellTypeTaxon : "contains_taxon"
 CellTypeTaxon ||--|o CellTypeTaxon : "has_parent"
 CellTypeTaxon ||--|o CellTypeTaxonomy : "part_of_taxonomy"
-CellTypeTaxon ||--|o ProvActivity : "was_generated_by"
-CellTypeTaxon ||--|o ProvEntity : "was_derived_from"
 CellTypeTaxon ||--|o SpatialProportions : "spatial_proportions_human, spatial_proportions_macaque, spatial_proportions_marmoset, spatial_regional_proportions"
 CellTypeTaxon ||--}o Abbreviation : "has_abbreviation"
-CellTypeTaxon ||--}o Attribute : "has attribute"
 CellTypeTaxon ||--}o Cluster : "contains_cluster"
 CellTypeTaxon ||--}o GeneAnnotation : "curated_markers_to_mouse, curated_markers_to_primates"
 CellTypeTaxonomy ||--|o CellTypeTaxonomy : "is_revision_of"
 CellTypeTaxonomy ||--|o CellTypeTaxonomyCreationProcess : "was_generated_by"
-CellTypeTaxonomy ||--}o Attribute : "has attribute"
 CellTypeTaxonomy ||--}o ClusterSet : "was_derived_from"
-CellTypeTaxonomyCreationProcess ||--}o Attribute : "has attribute"
 CellTypeTaxonomyCreationProcess ||--}o ClusterSet : "used"
 Cluster ||--|o ClusterSet : "part_of_set"
-Cluster ||--|o ProvActivity : "was_generated_by"
-Cluster ||--|o ProvEntity : "was_derived_from"
-Cluster ||--}o Attribute : "has attribute"
 Cluster ||--}o CellSpecimen : "contains_sample"
 Cluster ||--}o ObservationRow : "contains_observation"
 ClusterSet ||--|o ClusterSet : "is_revision_of"
 ClusterSet ||--|o ClusteringProcess : "was_generated_by"
-ClusterSet ||--}o Attribute : "has attribute"
 ClusterSet ||--}o ObservationMatrix : "was_derived_from"
-ClusteringProcess ||--}o Attribute : "has attribute"
 ClusteringProcess ||--}o ObservationMatrix : "used"
 ColorPalette ||--|o CellTypeTaxonomy : "is_palette_for"
-ColorPalette ||--|o ProvActivity : "was_generated_by"
-ColorPalette ||--|o ProvEntity : "was_derived_from"
-ColorPalette ||--}o Attribute : "has attribute"
 DisplayColor ||--|o CellTypeSet : "is_color_for_set"
 DisplayColor ||--|o CellTypeTaxon : "is_color_for_taxon"
 DisplayColor ||--|o ColorPalette : "part_of_palette"
-DisplayColor ||--|o ProvActivity : "was_generated_by"
-DisplayColor ||--|o ProvEntity : "was_derived_from"
-DisplayColor ||--}o Attribute : "has attribute"
-MatrixFile ||--|o ProvActivity : "was_generated_by"
-MatrixFile ||--|o ProvEntity : "was_derived_from"
-MatrixFile ||--}o Attribute : "has attribute"
 ObservationMatrix ||--|o ObservationMatrixCreationProcess : "was_generated_by"
-ObservationMatrix ||--}o Attribute : "has attribute"
 ObservationMatrix ||--}o CellSpecimen : "was_derived_from"
 ObservationMatrix ||--}o GeneAnnotation : "has_variable"
 ObservationMatrix ||--}o MatrixFile : "represented_by"
-ObservationMatrixCreationProcess ||--|o ProvEntity : "used"
-ObservationMatrixCreationProcess ||--}o Attribute : "has attribute"
 ObservationRow ||--|o CellSpecimen : "was_derived_from"
 ObservationRow ||--|o MatrixFile : "represented_in"
 ObservationRow ||--|o ObservationMatrix : "part_of_matrix"
-ObservationRow ||--|o ProvActivity : "was_generated_by"
-ObservationRow ||--}o Attribute : "has attribute"
-ProvActivity ||--|o ProvEntity : "used"
-ProvEntity ||--|o ProvActivity : "was_generated_by"
-ProvEntity ||--|o ProvEntity : "was_derived_from"
-SpatialProportions ||--|o ProvActivity : "was_generated_by"
-SpatialProportions ||--|o ProvEntity : "was_derived_from"
-SpatialProportions ||--}o Attribute : "has attribute"
 
 ```
-

erdiagram-autogen/cell_taxonomy.md

@@ -162,51 +162,19 @@ ExpressionMatrix {
     stringList type  
     uriorcurieList xref  
 }
-ProvActivity {
-
-}
-ProvEntity {
-
-}
 
 Cell ||--|o Cluster : "part_of_cluster"
-Cell ||--|o ProvActivity : "was_generated_by"
-Cell ||--|o ProvEntity : "was_derived_from"
-Cell ||--}o Attribute : "has attribute"
 CellTypeSet ||--|o CellTypeSet : "has_parent"
 CellTypeSet ||--|o CellTypeTaxonomy : "part_of_taxonomy"
-CellTypeSet ||--|o ProvActivity : "was_generated_by"
-CellTypeSet ||--|o ProvEntity : "was_derived_from"
-CellTypeSet ||--}o Attribute : "has attribute"
 CellTypeTaxon ||--|o CellTypeSet : "part_of_set"
 CellTypeTaxon ||--|o CellTypeTaxon : "has_parent"
-CellTypeTaxon ||--|o ProvActivity : "was_generated_by"
-CellTypeTaxon ||--|o ProvEntity : "was_derived_from"
-CellTypeTaxon ||--}o Attribute : "has attribute"
 CellTypeTaxon ||--}o GeneAnnotation : "curated_markers_to_mouse, curated_markers_to_primates"
-CellTypeTaxonomy ||--|o ProvActivity : "was_generated_by"
-CellTypeTaxonomy ||--}o Attribute : "has attribute"
 CellTypeTaxonomy ||--}o ClusterSet : "was_derived_from"
 CellTypeTaxonomy ||--}o Embedding : "has_embedding"
 CellTypeTaxonomy ||--}o ExpressionMatrix : "has_expression_matrix"
 Cluster ||--|o ClusterSet : "part_of_set"
-Cluster ||--|o ProvActivity : "was_generated_by"
-Cluster ||--|o ProvEntity : "was_derived_from"
-Cluster ||--}o Attribute : "has attribute"
 Cluster ||--}o CellTypeTaxon : "has_parent"
-ClusterSet ||--|o ProvActivity : "was_generated_by"
-ClusterSet ||--}o Attribute : "has attribute"
 ClusterSet ||--}o ExpressionMatrix : "was_derived_from"
-Embedding ||--|o ProvActivity : "was_generated_by"
-Embedding ||--|o ProvEntity : "was_derived_from"
-Embedding ||--}o Attribute : "has attribute"
-ExpressionMatrix ||--|o ProvActivity : "was_generated_by"
-ExpressionMatrix ||--|o ProvEntity : "was_derived_from"
-ExpressionMatrix ||--}o Attribute : "has attribute"
 ExpressionMatrix ||--}o GeneAnnotation : "has_variable"
-ProvActivity ||--|o ProvEntity : "used"
-ProvEntity ||--|o ProvActivity : "was_generated_by"
-ProvEntity ||--|o ProvEntity : "was_derived_from"
 
 ```
-

erdiagram-autogen/genome_annotation.md

@@ -72,12 +72,8 @@ GenomeAssembly {
 AnnotationCollection ||--}o GeneAnnotation : "annotations"
 AnnotationCollection ||--}o GenomeAnnotation : "genome_annotations"
 AnnotationCollection ||--}o GenomeAssembly : "genome_assemblies"
-GeneAnnotation ||--}o Attribute : "has attribute"
 GeneAnnotation ||--}o OrganismTaxon : "in taxon"
-GenomeAnnotation ||--}o Attribute : "has attribute"
 GenomeAnnotation ||--}o OrganismTaxon : "in taxon"
-GenomeAssembly ||--}o Attribute : "has attribute"
 GenomeAssembly ||--}o OrganismTaxon : "in taxon"
 
 ```
-

erdiagram-autogen/library_generation.md

@@ -341,12 +341,6 @@ LibraryPooling {
     label_typeList synonym  
     stringList type  
     uriorcurieList xref  
-}
-ProvActivity {
-
-}
-ProvEntity {
-
 }
 TissueDissection {
     string id  
@@ -384,65 +378,27 @@ TissueSample {
 
 AmplifiedCdna ||--|o BarcodedCellSample : "was_derived_from"
 AmplifiedCdna ||--|o CdnaAmplification : "was_generated_by"
-AmplifiedCdna ||--}o Attribute : "has attribute"
 BarcodedCellSample ||--|o CellBarcoding : "was_generated_by"
-BarcodedCellSample ||--}o Attribute : "has attribute"
-BarcodedCellSample ||--}o ProvEntity : "was_derived_from"
-BrainSlab ||--|o ProvActivity : "was_generated_by"
-BrainSlab ||--|o ProvEntity : "was_derived_from"
-BrainSlab ||--}o Attribute : "has attribute"
 CdnaAmplification ||--|o BarcodedCellSample : "used"
-CdnaAmplification ||--}o Attribute : "has attribute"
-CellBarcoding ||--}o Attribute : "has attribute"
-CellBarcoding ||--}o ProvEntity : "used"
-CellDissociation ||--}o Attribute : "has attribute"
 CellDissociation ||--}o TissueSample : "used"
-CellEnrichment ||--}o Attribute : "has attribute"
 CellEnrichment ||--}o DissociatedCellSample : "used"
 DigitalAsset ||--|o LibraryPool : "was_derived_from"
-DigitalAsset ||--|o ProvActivity : "was_generated_by"
-DigitalAsset ||--}o Attribute : "has attribute"
 DissectionRoiDelineation ||--|o BrainSlab : "used"
-DissectionRoiDelineation ||--}o Attribute : "has attribute"
 DissectionRoiPolygon ||--|o BrainSlab : "annotates"
 DissectionRoiPolygon ||--|o DissectionRoiDelineation : "was_generated_by"
-DissectionRoiPolygon ||--|o ProvEntity : "was_derived_from"
-DissectionRoiPolygon ||--}o Attribute : "has attribute"
 DissociatedCellSample ||--|o CellDissociation : "was_generated_by"
-DissociatedCellSample ||--}o Attribute : "has attribute"
 DissociatedCellSample ||--}o TissueSample : "was_derived_from"
-Donor ||--|o ProvActivity : "was_generated_by"
-Donor ||--|o ProvEntity : "was_derived_from"
-Donor ||--}o Attribute : "has attribute"
 Donor ||--}o OrganismTaxon : "in taxon"
-EnrichedCellSample ||--|o ProvActivity : "was_generated_by"
-EnrichedCellSample ||--}o Attribute : "has attribute"
-EnrichedCellSample ||--}o ProvEntity : "was_derived_from"
 EnrichedCellSampleSplitting ||--|o EnrichedCellSample : "used"
-EnrichedCellSampleSplitting ||--}o Attribute : "has attribute"
 Library ||--|o LibraryConstruction : "was_generated_by"
-Library ||--|o ProvEntity : "was_derived_from"
-Library ||--}o Attribute : "has attribute"
 LibraryAliquot ||--|o Library : "was_derived_from"
-LibraryAliquot ||--|o ProvActivity : "was_generated_by"
-LibraryAliquot ||--}o Attribute : "has attribute"
-LibraryConstruction ||--|o ProvEntity : "used"
-LibraryConstruction ||--}o Attribute : "has attribute"
 LibraryPool ||--|o LibraryPooling : "was_generated_by"
-LibraryPool ||--}o Attribute : "has attribute"
 LibraryPool ||--}o LibraryAliquot : "was_derived_from"
-LibraryPooling ||--}o Attribute : "has attribute"
 LibraryPooling ||--}o LibraryAliquot : "used"
-ProvActivity ||--|o ProvEntity : "used"
-ProvEntity ||--|o ProvActivity : "was_generated_by"
-ProvEntity ||--|o ProvEntity : "was_derived_from"
 TissueDissection ||--|o BrainSlab : "used"
 TissueDissection ||--|o DissectionRoiPolygon : "was_guided_by"
-TissueDissection ||--}o Attribute : "has attribute"
 TissueSample ||--|o DissectionRoiPolygon : "dissection_was_guided_by"
 TissueSample ||--|o Donor : "was_derived_from"
 TissueSample ||--|o TissueDissection : "was_generated_by"
-TissueSample ||--}o Attribute : "has attribute"
 
 ```
-

utils/fix_and_create_erdiagram.py

@@ -2,6 +2,26 @@
 import subprocess as sp
 import yaml
 
+# Classes inherited from bican_biolink that add noise without domain value.
+CLASSES_TO_EXCLUDE = {"Attribute", "ProvActivity", "ProvEntity"}
+
+
+def _remove_excluded_classes(diagram_txt: str) -> str:
+    lines = diagram_txt.splitlines()
+    result = []
+    in_excluded_block = False
+    for line in lines:
+        if any(line.startswith(f"{cls} {{") for cls in CLASSES_TO_EXCLUDE):
+            in_excluded_block = True
+        if in_excluded_block:
+            if line.strip() == "}":
+                in_excluded_block = False
+            continue
+        if any(f" {cls} : " in line for cls in CLASSES_TO_EXCLUDE):
+            continue
+        result.append(line)
+    return "\n".join(result)
+
 
 def fix_diagram(yaml_model: pathlib.Path, diagram: pathlib.Path):
     # reading the yaml schema and getting all names of classes
@@ -22,6 +42,9 @@ def fix_diagram(yaml_model: pathlib.Path, diagram: pathlib.Path):
     with diagram.open() as f:
         diagram_txt = f.read()
 
+    # remove classes inherited from bican_biolink that add noise to domain diagrams
+    diagram_txt = _remove_excluded_classes(diagram_txt)
+
     to_fix_list = [
         "iri type",
         "predicate type",