[POST-berkeley-MERGER] Update FunctionalAnnotationAggMember class for compatibility with MetaP Aggregation tables

nmdc_schema/migrators/migrator_from_XX_to_PR242.py

@@ -0,0 +1,32 @@
+from nmdc_schema.migrators.migrator_base import MigratorBase
+
+
+class Migrator(MigratorBase):
+    r"""Migrates a database between two schemas."""
+
+    _from_version = "XX"
+    _to_version = "XX" # PR242
+
+    def upgrade(self):
+        r"""Migrates the database from conforming to the original schema, to conforming to the new schema."""
+
+        self.adapter.process_each_document(
+            "functional_annotation_agg", [self.move_metagenome_id_to_was_generated_by]
+        )
+
+    def move_metagenome_id_to_was_generated_by(self, fun_agg: dict) -> dict:
+        r"""
+        Updates the `FunctionalAnnotationAggMember` records so the value originally in its `metagenome_annotation_id` field 
+        is stored in a new field named `was_generated_by`; and removes the `metagenome_annotation_id` field.
+
+        `metagenome_annotation_id` is required on these records and has the same value as `was_generated_by` in the new schema,
+        so no data is lost in this migration nor do we need to check for the existence of the field.
+
+        >>> m = Migrator()
+        >>> m.move_metagenome_id_to_was_generated_by({'id': 123, 'metagenome_annotation_id': 'mgm123', 'count': 1})
+        {'id': 123, 'count': 1, 'was_generated_by': 'mgm123'}
+
+        """
+        self.logger.info(f"Migrating Extraction: {fun_agg['id']}")
+        fun_agg["was_generated_by"] = fun_agg.pop("metagenome_annotation_id")
+        return fun_agg

src/data/invalid/Database-functional_annotation_agg.yaml

@@ -1,4 +1,4 @@
 #invalid b/c a required slot is missing
 functional_annotation_agg:
-  - metagenome_annotation_id: nmdc:8253bcdcd0387177ff895c38a047c719
+  - was_generated_by: nmdc:8253bcdcd0387177ff895c38a047c719
     count: 120

src/data/invalid/FunctionalAnnotationAggMember-invalid-no_count.yaml

@@ -0,0 +1,4 @@
+# This example is invalid because the count field is missing and the was_generated_by field refers to a MetagenomeAnnotation
+type: nmdc:FunctionalAnnotationAggMember
+was_generated_by: nmdc:wfmgan-99-123456.1
+gene_function_id: KEGG.ORTHOLOGY:K00627

src/data/valid/Database-functional_annotation_agg.yaml

@@ -1,5 +1,5 @@
 functional_annotation_agg:
-  - metagenome_annotation_id: nmdc:wfmgan-99-123456.1
+  - was_generated_by: nmdc:wfmgan-99-123456.1
     gene_function_id: KEGG.ORTHOLOGY:K00627
     count: 120
     type: nmdc:FunctionalAnnotationAggMember

src/data/valid/Database-interleaved.yaml

@@ -426,7 +426,7 @@ functional_annotation_agg:
   - gene_function_id: KEGG.ORTHOLOGY:K00627
     count: 120
     type: nmdc:FunctionalAnnotationAggMember
-    metagenome_annotation_id: nmdc:wfmgan-99-123456.1
+    was_generated_by: nmdc:wfmgan-99-123456.1
 biosample_set:
   - id: nmdc:bsm-99-isqhuW
     type: nmdc:Biosample

src/data/valid/FunctionalAnnotationAggMember-minimal.yaml

@@ -1,4 +1,4 @@
-metagenome_annotation_id: nmdc:wfmgan-99-123456.1
+was_generated_by: nmdc:wfmgan-99-123456.1
 gene_function_id: KEGG.ORTHOLOGY:K00627
 count: 120
 type: nmdc:FunctionalAnnotationAggMember

src/schema/nmdc.yaml

@@ -242,15 +242,28 @@ classes:
   FunctionalAnnotationAggMember:
     class_uri: nmdc:FunctionalAnnotationAggMember
     slots:
-      - metagenome_annotation_id
+      - was_generated_by
       - gene_function_id
       - count
       - type
     slot_usage:
-      metagenome_annotation_id:
+      was_generated_by:
         structured_pattern:
-          syntax: "{id_nmdc_prefix}:wfmgan-{id_shoulder}-{id_blade}{id_version}{id_locus}$"
+          syntax: "{id_nmdc_prefix}:(wfmgan|wfmp)-{id_shoulder}-{id_blade}{id_version}$"
           interpolated: true
+        required: true
+    rules:
+      - title: count_required_if_MetagenomeAnnotation
+        description: >-
+          If was_generated_by is a MetagenomeAnnotation, then count is required.
+        preconditions:
+          slot_conditions:
+            was_generated_by:
+              pattern: ^.+:wfmgan-.+$
+        postconditions:
+          slot_conditions:
+            count:
+              required: true
 
   Database:
     class_uri: nmdc:Database
@@ -925,6 +938,7 @@ slots:
     range: WorkflowExecution
     description: The identifier for the analysis activity that generated the functional annotation results, where the analysis activity is an instance of the/an appropriate subclass of WorkflowExecution
     required: true
+    deprecated: "not used. 2024-10 https://github.com/microbiomedata/nmdc-schema/issues/1253"
 
   gene_function_id:
     range: uriorcurie
@@ -934,7 +948,6 @@ slots:
     required: true
   count:
     range: integer
-    required: true
   functional_annotation_agg:
     range: FunctionalAnnotationAggMember
     multivalued: true