[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/FunctionalAnnotationAggMember-invalid-no_count.yaml

@@ -0,0 +1,4 @@
+# This example is invalid because the count field is and it is required.
+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-metatranscriptome.yaml

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

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

@@ -255,16 +255,23 @@ classes:
 
   FunctionalAnnotationAggMember:
     class_uri: nmdc:FunctionalAnnotationAggMember
+    unique_keys: 
+      main:
+        description: A FunctionalAnntoationAggMember record is uniquely identified by the combination of the was_generated_by and gene_function_id slots.
+        unique_key_slots:
+          - was_generated_by
+          - gene_function_id
     slots:
-      - metagenome_annotation_id
+      - was_generated_by
       - gene_function_id
       - count
       - type
     slot_usage:
-      metagenome_annotation_id:
-        structured_pattern: # doesn't include act
-          syntax: "{id_nmdc_prefix}:(wfmgan|wfmtan)-{id_shoulder}-{id_blade}{id_version}$"
+      was_generated_by:
+        structured_pattern:
+          syntax: "{id_nmdc_prefix}:(wfmgan|wfmp|wfmtan)-{id_shoulder}-{id_blade}{id_version}$"
           interpolated: true
+        required: true
 
   Database:
     class_uri: nmdc:Database
@@ -958,19 +965,21 @@ 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
-    any_of:
-      - range: MetagenomeAnnotation
-      - range: MetatranscriptomeAnnotation
+    deprecated: "not used. 2024-10 https://github.com/microbiomedata/nmdc-schema/issues/1253"
+
 
   gene_function_id:
     range: uriorcurie
     description: The identifier for the gene function.
     examples:
       - value: KEGG.ORTHOLOGY:K00627
     required: true
+
   count:
     range: integer
+    description: The number of sequences (for a metagenome or metatranscriptome) or spectra (for metaproteomics) associated with the specified function.
     required: true
+
   functional_annotation_agg:
     range: FunctionalAnnotationAggMember
     multivalued: true