chore: Updating configs and validation (#244)

* Updating configs and validation * reorganize annotationmethodlinks * formatting --------- Co-authored-by: Daniel Ji <[email protected]>
chanzuckerberg · Sep 3, 2024 · 6c9f95c · 6c9f95c
1 parent cd54d91
commit 6c9f95c
Show file tree

Hide file tree

Showing 18 changed files with 447 additions and 397 deletions.
diff --git a/apiv2/schema/schema.yaml b/apiv2/schema/schema.yaml
@@ -2552,7 +2552,7 @@ classes:
         range: boolean
   AnnotationMethodLinks:
     name: AnnotationMethodLinks
-    description: A set of links to models, sourcecode, documentation, etc referenced
+    description: A set of links to models, source code, documentation, etc referenced
       by annotation the method
     mixin: true
     from_schema: cdp-dataset-config
@@ -2565,7 +2565,7 @@ classes:
         required: true
       link_type:
         name: link_type
-        description: Type of link (e.g. model, sourcecode, documentation)
+        description: Type of link (e.g. model, source code, documentation)
         from_schema: cdp-dataset-config
         range: annotation_method_link_type_enum
         required: true

diff --git a/ingestion_tools/dataset_configs/template.yaml b/ingestion_tools/dataset_configs/template.yaml
@@ -32,7 +32,7 @@ annotations: OPTIONAL
     method_links: OPTIONAL
     - link: REQUIRED, STRING
       link_type: REQUIRED, STRING (ENUM [documentation, model_weights, other, source_code, website])
-      name: OPTIONAL, STRING
+      custom_name: OPTIONAL, STRING
     version: OPTIONAL, FLOAT (NON-NEGATIVE)
   sources: REQUIRED
   # The following are permissible source types (multiple, but different types can exist per annotation,

diff --git a/schema/api/v1.0.0/api_models.yaml b/schema/api/v1.0.0/api_models.yaml
@@ -162,6 +162,9 @@ classes:
       method_type:
         exact_mappings:
           - cdp-common:annotation_method_type
+      method_links:
+        exact_mappings:
+          - cdp-common:annotation_method_links_json
     annotations:
       plural: Annotations
   DatasetAuthor:

diff --git a/schema/api/v1.0.0/codegen/ER_DIAGRAM.md b/schema/api/v1.0.0/codegen/ER_DIAGRAM.md
@@ -72,6 +72,7 @@ Annotation {
     string annotation_software
     boolean is_curator_recommended
     annotation_method_type_enum method_type
+    string method_links
     integer id
     date deposition_date
     date release_date

diff --git a/schema/api/v1.0.0/codegen/api_models_materialized.yaml b/schema/api/v1.0.0/codegen/api_models_materialized.yaml
@@ -664,7 +664,6 @@ classes:
         owner: AuthorEntityMixin
         domain_of:
         - AuthorMixin
-        - AnnotationMethodLinks
         - DatasetAuthor
         - Run
         - Tomograms
@@ -839,7 +838,6 @@ classes:
         owner: AnnotationAuthor
         domain_of:
         - AuthorMixin
-        - AnnotationMethodLinks
         - DatasetAuthor
         - Run
         - Tomograms
@@ -1348,6 +1346,20 @@ classes:
         inlined: true
         inlined_as_list: true
         pattern: (^manual$)|(^automated$)|(^hybrid$)|(^simulated$)
+      method_links:
+        name: method_links
+        description: A set of links to models, source code, documentation, etc referenced
+          by annotation the method as json
+        from_schema: cdp-api-models
+        exact_mappings:
+        - cdp-common:annotation_method_links_json
+        alias: method_links
+        owner: Annotation
+        domain_of:
+        - Annotation
+        range: string
+        inlined: true
+        inlined_as_list: true
       id:
         name: id
         description: An identifier to refer to a specific instance of this type
@@ -1455,7 +1467,6 @@ classes:
         owner: DatasetAuthor
         domain_of:
         - AuthorMixin
-        - AnnotationMethodLinks
         - DatasetAuthor
         - Run
         - Tomograms
@@ -2250,7 +2261,6 @@ classes:
         owner: Run
         domain_of:
         - AuthorMixin
-        - AnnotationMethodLinks
         - DatasetAuthor
         - Run
         - Tomograms
@@ -2943,7 +2953,6 @@ classes:
         owner: TomogramAuthor
         domain_of:
         - AuthorMixin
-        - AnnotationMethodLinks
         - DatasetAuthor
         - Run
         - Tomograms
@@ -3215,7 +3224,6 @@ classes:
         owner: Tomograms
         domain_of:
         - AuthorMixin
-        - AnnotationMethodLinks
         - DatasetAuthor
         - Run
         - Tomograms
@@ -3741,7 +3749,6 @@ classes:
         owner: AuthorMixin
         domain_of:
         - AuthorMixin
-        - AnnotationMethodLinks
         - DatasetAuthor
         - Run
         - Tomograms
@@ -3836,51 +3843,4 @@ classes:
         range: boolean
         inlined: true
         inlined_as_list: true
-  AnnotationMethodLinks:
-    name: AnnotationMethodLinks
-    description: A set of links to models, sourcecode, documentation, etc referenced
-      by annotation the method
-    from_schema: cdp-api-models
-    attributes:
-      link:
-        name: link
-        description: URL to the resource
-        from_schema: cdp-api-models
-        alias: link
-        owner: AnnotationMethodLinks
-        domain_of:
-        - AnnotationMethodLinks
-        range: string
-        required: true
-        inlined: true
-        inlined_as_list: true
-      link_type:
-        name: link_type
-        description: Type of link (e.g. model, sourcecode, documentation)
-        from_schema: cdp-api-models
-        alias: link_type
-        owner: AnnotationMethodLinks
-        domain_of:
-        - AnnotationMethodLinks
-        range: annotation_method_link_type_enum
-        required: true
-        inlined: true
-        inlined_as_list: true
-        pattern: (^documentation$)|(^models_weights$)|(^other$)|(^source_code$)|(^website$)
-      name:
-        name: name
-        description: user readable name of the resource
-        from_schema: cdp-api-models
-        alias: name
-        owner: AnnotationMethodLinks
-        domain_of:
-        - AuthorMixin
-        - AnnotationMethodLinks
-        - DatasetAuthor
-        - Run
-        - Tomograms
-        range: string
-        recommended: true
-        inlined: true
-        inlined_as_list: true
 source_file: api/v1.0.0/api_models.yaml
diff --git a/schema/api/v2.0.0/api_models.yaml b/schema/api/v2.0.0/api_models.yaml
@@ -254,6 +254,9 @@ classes:
       method_type:
         exact_mappings:
           - cdp-common:annotation_method_type
+      method_links:
+        exact_mappings:
+          - cdp-common:annotation_method_links_json
     annotations:
       plural: Annotations
   DatasetAuthor:

diff --git a/schema/api/v2.0.0/codegen/ER_DIAGRAM.md b/schema/api/v2.0.0/codegen/ER_DIAGRAM.md
@@ -51,8 +51,8 @@ Tomogram {
     boolean is_canonical
     string s3_omezarr_dir
     string https_omezarr_dir
-    string s3_mrc_scale0
-    string https_mrc_scale0
+    string s3_mrc_file
+    string https_mrc_file
     string scale0_dimensions
     string scale1_dimensions
     string scale2_dimensions
@@ -96,7 +96,7 @@ Annotation {
     EMPIAR_EMDB_DOI_PDB_LIST annotation_publication
     string annotation_method
     boolean ground_truth_status
-    GO_ID object_id
+    string object_id
     string object_name
     string object_description
     string object_state
@@ -107,6 +107,7 @@ Annotation {
     string annotation_software
     boolean is_curator_recommended
     annotation_method_type_enum method_type
+    string method_links
     integer id
     date deposition_date
     date release_date
@@ -131,9 +132,9 @@ DepositionType {
 }
 Tiltseries {
     string s3_omezarr_dir
-    string s3_mrc_bin1
+    string s3_mrc_file
     string https_omezarr_dir
-    string https_mrc_bin1
+    string https_mrc_file
     string s3_collection_metadata
     string https_collection_metadata
     string s3_angle_list
@@ -217,8 +218,8 @@ DepositionAuthor {
     string affiliation_name
     string affiliation_address
     string affiliation_identifier
-    string corresponding_author_status
-    string primary_author_status
+    boolean corresponding_author_status
+    boolean primary_author_status
 }
 AnnotationAuthor {
     integer id
@@ -229,8 +230,8 @@ AnnotationAuthor {
     string affiliation_name
     string affiliation_address
     string affiliation_identifier
-    string corresponding_author_status
-    string primary_author_status
+    boolean corresponding_author_status
+    boolean primary_author_status
 }
 TomogramAuthor {
     integer id
@@ -241,8 +242,8 @@ TomogramAuthor {
     string affiliation_name
     string affiliation_address
     string affiliation_identifier
-    string corresponding_author_status
-    string primary_author_status
+    boolean corresponding_author_status
+    boolean primary_author_status
 }
 DatasetAuthor {
     integer id
@@ -253,8 +254,8 @@ DatasetAuthor {
     string affiliation_name
     string affiliation_address
     string affiliation_identifier
-    string corresponding_author_status
-    string primary_author_status
+    boolean corresponding_author_status
+    boolean primary_author_status
 }
 DatasetFunding {
     string funding_agency_name

diff --git a/schema/api/v2.0.0/codegen/api_models_materialized.yaml b/schema/api/v2.0.0/codegen/api_models_materialized.yaml
@@ -1764,6 +1764,20 @@ classes:
         inlined: true
         inlined_as_list: true
         pattern: (^manual$)|(^automated$)|(^hybrid$)|(^simulated$)
+      method_links:
+        name: method_links
+        description: A set of links to models, source code, documentation, etc referenced
+          by annotation the method as json
+        from_schema: cdp-api
+        exact_mappings:
+        - cdp-common:annotation_method_links_json
+        alias: method_links
+        owner: Annotation
+        domain_of:
+        - Annotation
+        range: string
+        inlined: true
+        inlined_as_list: true
       id:
         name: id
         description: An identifier to refer to a specific instance of this type
@@ -3675,16 +3689,17 @@ classes:
         range: string
         inlined: true
         inlined_as_list: true
-      s3_mrc_bin1:
-        name: s3_mrc_bin1
+      s3_mrc_file:
+        name: s3_mrc_file
         description: S3 path to this tiltseries in MRC format (no scaling)
         from_schema: cdp-api
         exact_mappings:
-        - cdp-common:tiltseries_s3_mrc_bin1
-        alias: s3_mrc_bin1
+        - cdp-common:tiltseries_s3_mrc_file
+        alias: s3_mrc_file
         owner: Tiltseries
         domain_of:
         - Tiltseries
+        - Tomogram
         range: string
         inlined: true
         inlined_as_list: true
@@ -3702,16 +3717,17 @@ classes:
         range: string
         inlined: true
         inlined_as_list: true
-      https_mrc_bin1:
-        name: https_mrc_bin1
+      https_mrc_file:
+        name: https_mrc_file
         description: HTTPS path to this tiltseries in MRC format (no scaling)
         from_schema: cdp-api
         exact_mappings:
-        - cdp-common:tiltseries_https_mrc_bin1
-        alias: https_mrc_bin1
+        - cdp-common:tiltseries_https_mrc_file
+        alias: https_mrc_file
         owner: Tiltseries
         domain_of:
         - Tiltseries
+        - Tomogram
         range: string
         inlined: true
         inlined_as_list: true
@@ -4779,28 +4795,30 @@ classes:
         range: string
         inlined: true
         inlined_as_list: true
-      s3_mrc_scale0:
-        name: s3_mrc_scale0
+      s3_mrc_file:
+        name: s3_mrc_file
         description: S3 path to this tomogram in MRC format (no scaling)
         from_schema: cdp-api
         exact_mappings:
-        - cdp-common:tomogram_s3_mrc_scale0
-        alias: s3_mrc_scale0
+        - cdp-common:tomogram_s3_mrc_file
+        alias: s3_mrc_file
         owner: Tomogram
         domain_of:
+        - Tiltseries
         - Tomogram
         range: string
         inlined: true
         inlined_as_list: true
-      https_mrc_scale0:
-        name: https_mrc_scale0
+      https_mrc_file:
+        name: https_mrc_file
         description: HTTPS path to this tomogram in MRC format (no scaling)
         from_schema: cdp-api
         exact_mappings:
-        - cdp-common:tomogram_https_mrc_scale0
-        alias: https_mrc_scale0
+        - cdp-common:tomogram_https_mrc_file
+        alias: https_mrc_file
         owner: Tomogram
         domain_of:
+        - Tiltseries
         - Tomogram
         range: string
         inlined: true