rearrange enumerations

content/linkml/AnVILBioCoreMinimal.linkml.yaml

@@ -45,8 +45,9 @@ classes:
       - reference_assembly
       - biosample_id_fk # this is the foregin key to the biosample table
 
-slots:
-# ids are unique identifiers for each entity or explicitly defined foreign key
+slots: ## ids are unique identifiers for each entity or explicitly defined foreign key
+
+# foreign keys
   donor_id_fk:
     aliases: hasDonor
     description: This property references the Donor organism from which the BioSample was acquired.
@@ -61,6 +62,8 @@ slots:
     domain: AnVILFile
     multivalued: false # assume that a biosample is only from one donor (1-1 relationship) 
     inlined: true
+
+  # primary keys
   biosample_id:
     identifier: true  # this is the type of primary key
     range: string
@@ -103,12 +106,6 @@ slots:
     comments:
     - 'For example: Homo sapiens from NCBITaxon or http://purl.obolibrary.org/obo/NCBITaxon_9606 '
     multivalued: false
-  phenotypic_sex:
-    aliases: hasPhenotypicSex
-    description: A reference to the BiologicalSex of the Donor organism. \"An organismal quality inhering in a bearer by virtue of the bearer's physical expression of sexual characteristics. [PATO_0001894]\
-    multivalued: false
-    inlined: true
-    range: PhenotypicSex
   reported_ethnicity:
     aliases: hasReportedEthnicity
     description: A property that relects a Human Donor's reported ethnic origins. Note this may contain both Race and Ethnicity information as define by the US Department of Interior (DOI) https://www.doi.gov/pmb/eeo/directives/race-data
@@ -147,21 +144,29 @@ slots:
     aliases: usesReferenceAssembly
     description: A reference to the collection of sequences taken as the standard for a given organism. May be defined by https://www.ncbi.nlm.nih.gov/grc.
     multivalued: true
+
+  # fields for enumerations
   human_phenotype:
     description: phenotype code from the human phenotype ontology (HPO)
     # range: HumanPhenotypeOntology
+  phenotypic_sex:
+    aliases: hasPhenotypicSex
+    description: A reference to the BiologicalSex of the Donor organism. \"An organismal quality inhering in a bearer by virtue of the bearer's physical expression of sexual characteristics. [PATO_0001894]\
+    multivalued: false
+    inlined: true
+    # range: PhenotypicSex
 
-enums:
-  # HumanPhenotypeOntology:
-  #   permissible_values:
-  #     All:
-  PhenotypicSex:
-    permissible_values:
-      female:
-        meaning: PATO:0000383
-      male:
-        meaning: PATO:0000384
-      intersex:
-        description: between male and female sex
-      prefer not to say:
-        description: the individual prefers not to disclose
+# enums:
+#   # HumanPhenotypeOntology:
+#   #   permissible_values:
+#   #     All:
+#   PhenotypicSex:
+#     permissible_values:
+#       female:
+#         meaning: PATO:0000383
+#       male:
+#         meaning: PATO:0000384
+#       intersex:
+#         description: between male and female sex
+#       prefer not to say:
+#         description: the individual prefers not to disclose

content/tools/python/input/AnVILBioCoreMinimal.schema.json

@@ -64,8 +64,8 @@
                     "type": "array"
                 },
                 "human_phenotype": {
-                    "$ref": "#/$defs/HumanPhenotypeOntology",
-                    "description": "phenotype code from the human phenotype ontology (HPO)"
+                    "description": "phenotype code from the human phenotype ontology (HPO)",
+                    "type": "string"
                 },
                 "organism_type": {
                     "description": "A reference to the organism type.",
@@ -93,6 +93,10 @@
             "additionalProperties": false,
             "description": "Information for files associated with the study.",
             "properties": {
+                "biosample_id_fk": {
+                    "$ref": "#/$defs/AnVILBioSample",
+                    "description": "This property references the BioSample from which information in a file was acquired."
+                },
                 "data_modality": {
                     "description": "Data modality describes the biological nature of the information gathered as the result of an Activity, independent of the technology or methods used to produce the information.",
                     "items": {
@@ -136,14 +140,6 @@
             ],
             "title": "AnVILFile",
             "type": "object"
-        },
-        "HumanPhenotypeOntology": {
-            "description": "",
-            "enum": [
-                "All"
-            ],
-            "title": "HumanPhenotypeOntology",
-            "type": "string"
         }
     },
     "$id": "https://github.com/DataBiosphere/biocore-data-model/tree/main/content",

content/tools/python/output/biocore_tdr_schema.json

@@ -1 +1 @@
-{"tables": [{"name": "AnVILBioSample", "columns": [{"name": "anatomical_site", "datatype": "string", "array_of": false, "required": false}, {"name": "apriori_cell_type", "datatype": "string", "array_of": true, "required": false}, {"name": "biosample_id", "datatype": "string", "array_of": false, "required": true}, {"name": "biosample_type", "datatype": "string", "array_of": false, "required": false}, {"name": "disease", "datatype": "string", "array_of": false, "required": false}, {"name": "donor_age_at_collection_lower_bound", "datatype": "numeric", "array_of": false, "required": false}, {"name": "donor_age_at_collection_unit", "datatype": "string", "array_of": false, "required": false}, {"name": "donor_age_at_collection_upper_bound", "datatype": "numeric", "array_of": false, "required": false}, {"name": "donor_id", "datatype": "string", "array_of": false, "required": false}], "primaryKey": ["biosample_id"]}, {"name": "AnVILDonor", "columns": [{"name": "donor_id", "datatype": "string", "array_of": false, "required": true}, {"name": "genetic_ancestry", "datatype": "string", "array_of": true, "required": false}, {"name": "human_phenotype", "datatype": "string", "array_of": false, "required": false}, {"name": "organism_type", "datatype": "string", "array_of": false, "required": false}, {"name": "phenotypic_sex", "datatype": "string", "array_of": false, "required": false}, {"name": "reported_ethnicity", "datatype": "string", "array_of": true, "required": false}], "primaryKey": ["donor_id"]}, {"name": "AnVILFile", "columns": [{"name": "data_modality", "datatype": "string", "array_of": true, "required": false}, {"name": "file_format", "datatype": "string", "array_of": false, "required": false}, {"name": "file_id", "datatype": "string", "array_of": false, "required": true}, {"name": "file_md5sum", "datatype": "string", "array_of": false, "required": false}, {"name": "file_name", "datatype": "string", "array_of": false, "required": false}, {"name": "file_ref", "datatype": "string", "array_of": false, "required": false}, {"name": "file_size", "datatype": "integer", "array_of": false, "required": false}, {"name": "reference_assembly", "datatype": "string", "array_of": true, "required": false}], "primaryKey": ["file_id"]}], "relationships": [{"name": "AnVILBioSample.donor_id_to_AnVILDonor.donor_id", "from": {"table": "AnVILBioSample", "column": "donor_id"}, "to": {"table": "AnVILDonor", "column": "donor_id"}}]}
+{"tables": [{"name": "AnVILBioSample", "columns": [{"name": "anatomical_site", "datatype": "string", "array_of": false, "required": false}, {"name": "apriori_cell_type", "datatype": "string", "array_of": true, "required": false}, {"name": "biosample_id", "datatype": "string", "array_of": false, "required": true}, {"name": "biosample_type", "datatype": "string", "array_of": false, "required": false}, {"name": "disease", "datatype": "string", "array_of": false, "required": false}, {"name": "donor_age_at_collection_lower_bound", "datatype": "numeric", "array_of": false, "required": false}, {"name": "donor_age_at_collection_unit", "datatype": "string", "array_of": false, "required": false}, {"name": "donor_age_at_collection_upper_bound", "datatype": "numeric", "array_of": false, "required": false}, {"name": "donor_id", "datatype": "string", "array_of": false, "required": false}], "primaryKey": ["biosample_id"]}, {"name": "AnVILDonor", "columns": [{"name": "donor_id", "datatype": "string", "array_of": false, "required": true}, {"name": "genetic_ancestry", "datatype": "string", "array_of": true, "required": false}, {"name": "human_phenotype", "datatype": "string", "array_of": false, "required": false}, {"name": "organism_type", "datatype": "string", "array_of": false, "required": false}, {"name": "phenotypic_sex", "datatype": "string", "array_of": false, "required": false}, {"name": "reported_ethnicity", "datatype": "string", "array_of": true, "required": false}], "primaryKey": ["donor_id"]}, {"name": "AnVILFile", "columns": [{"name": "biosample_id", "datatype": "string", "array_of": false, "required": false}, {"name": "data_modality", "datatype": "string", "array_of": true, "required": false}, {"name": "file_format", "datatype": "string", "array_of": false, "required": false}, {"name": "file_id", "datatype": "string", "array_of": false, "required": true}, {"name": "file_md5sum", "datatype": "string", "array_of": false, "required": false}, {"name": "file_name", "datatype": "string", "array_of": false, "required": false}, {"name": "file_ref", "datatype": "string", "array_of": false, "required": false}, {"name": "file_size", "datatype": "integer", "array_of": false, "required": false}, {"name": "reference_assembly", "datatype": "string", "array_of": true, "required": false}], "primaryKey": ["file_id"]}], "relationships": [{"name": "AnVILBioSample.donor_id_to_AnVILDonor.donor_id", "from": {"table": "AnVILBioSample", "column": "donor_id"}, "to": {"table": "AnVILDonor", "column": "donor_id"}}, {"name": "AnVILFile.biosample_id_to_AnVILBioSample.biosample_id", "from": {"table": "AnVILFile", "column": "biosample_id"}, "to": {"table": "AnVILBioSample", "column": "biosample_id"}}]}