Merge pull request #2332 from microbiomedata/2330_metabolomics_category

Add slot and enumeration to `MetabolomicsAnalysis` and implements migrator

nmdc_schema/migrators/migrator_from_11_3_0_to_11_4_0.py

@@ -0,0 +1,35 @@
+from nmdc_schema.migrators.migrator_base import MigratorBase
+
+
+class Migrator(MigratorBase):
+    r"""Migrates a database between two schemas."""
+
+    _from_version = "11.3.0"
+    _to_version = "11.4.0"
+
+    def upgrade(self):
+        r"""Migrates the database from conforming to the original schema, to conforming to the new schema."""
+
+        self.adapter.process_each_document("workflow_execution_set", [self.set_metab_analysis_category])
+
+    def set_metab_analysis_category(self, workflow: dict) -> dict:
+        r"""
+        If the workflow execution record is of the type "nmdc:MetabolomicsAnalysis" and it has a `has_metabolite_identifications` field,
+        add field `metabolomics_analysis_category` and assign it the value "gc_ms_metabolomics". If the record does not 
+        have a `has_metabolite_identifications` field, it is assigned the value "lc_ms_lipidomics".
+
+        >>> m = Migrator()
+        >>> m.set_metab_analysis_category({'id': 123, 'type': 'nmdc:MetabolomicsAnalysis', 'has_metabolite_identifications': []})
+        {'id': 123, 'type': 'nmdc:MetabolomicsAnalysis', 'has_metabolite_identifications': [], 'metabolomics_analysis_category': 'gc_ms_metabolomics'}
+        >>> m.set_metab_analysis_category({'id': 123, 'type': 'nmdc:MetabolomicsAnalysis'})  # does not have has_metabolite_identifications field, therefore it's a lipid analysis
+        {'id': 123, 'type': 'nmdc:MetabolomicsAnalysis', 'metabolomics_analysis_category': 'lc_ms_lipidomics'}
+        >>> m.set_metab_analysis_category({'id': 123, 'type': 'nmdc:Metaproteomics'})  # not a metabolomics analysis
+        {'id': 123, 'type': 'nmdc:Metaproteomics'}
+        """
+
+        if workflow["type"] == "nmdc:MetabolomicsAnalysis":
+            if "has_metabolite_identifications" in workflow:
+                workflow["metabolomics_analysis_category"] = "gc_ms_metabolomics"
+            else:
+                workflow["metabolomics_analysis_category"] = "lc_ms_lipidomics"
+        return workflow

src/data/invalid/Database-MetabolomicsAnalysis-no_metabolomics_category.yaml

@@ -0,0 +1,16 @@
+# this example is invalid because the records in the workflow execution set do not have a metabolomics_analysis_category value
+workflow_execution_set:
+  - id: nmdc:wfmb-99-ABCDEF.1
+    name: Metabolomics Analysis Activity for nmdc:wfmb-99-ABCDEF.1
+    started_at_time: '2021-08-05T14:48:51+00:00'
+    ended_at_time: '2021-09-15T10:13:20+00:00'
+    execution_resource: NERSC-Cori
+    was_informed_by: nmdc:dgms-11-djad84
+    git_url: https://example.org/WorkflowExecutionActivity
+    has_input:
+      - nmdc:dobj-99-xxxxxx1
+      - nmdc:dobj-99-xxxxxx2
+    has_output:
+      - nmdc:dobj-99-xxxxxx3
+      - nmdc:dobj-99-xxxxxx4
+    type: nmdc:MetabolomicsAnalysis

src/data/invalid/Database-invalid_calibration_slot.yaml

@@ -52,3 +52,4 @@ workflow_execution_set:
   git_url: https://github.com/microbiomedata/metaMS
   started_at_time: '2021-01-07T23:54:40Z'
   was_informed_by: nmdc:dgms-13-122e4240
+  metabolomics_analysis_category: gc_ms_metabolomics

src/data/invalid/MetabolomicsAnalysis-invalid-has-slot-used.yaml

@@ -12,3 +12,4 @@ has_input:
 has_output:
   - nmdc:dobj-11-547rwa36
   - nmdc:dobj-11-547rwa37
+metabolomics_analysis_category: gc_ms_metabolomics

src/data/invalid/MetabolomicsAnalysis-invalid-missing_execution_resource.yaml

@@ -9,4 +9,5 @@ has_input:
 has_output:
   - nmdc:dobj-11-123d
   - nmdc:dobj-11-34dj4
-was_informed_by: nmdc:omprc-00-123456
+was_informed_by: nmdc:omprc-00-123456
+metabolomics_analysis_category: gc_ms_metabolomics

src/data/invalid/MetabolomicsAnalysis-invalid_execution_resource_not_in_enum.yaml

@@ -11,3 +11,4 @@ has_output:
   - nmdc:dobj-11-123d
   - nmdc:dobj-11-34dj4
 was_informed_by: nmdc:omprc-11-284u7d
+metabolomics_analysis_category: gc_ms_metabolomics

src/data/invalid/MetabolomicsAnalysis-invalid_id-1.yaml

@@ -12,3 +12,4 @@ has_output:
 - nmdc:dobj-11-ndgg7b37
 - nmdc:dobj-11-ndgg7b38
 type: nmdc:MetabolomicsAnalysis
+metabolomics_analysis_category: gc_ms_metabolomics

src/data/invalid/MetabolomicsAnalysis-invalid_id-2.yaml

@@ -12,3 +12,4 @@ has_output:
 - nmdc:dobj-11-ndgg7b37
 - nmdc:dobj-11-ndgg7b38
 type: nmdc:MetabolomicsAnalysis
+metabolomics_analysis_category: gc_ms_metabolomics

src/data/invalid/MetabolomicsAnalysis-invalid_id-3.yaml

@@ -12,3 +12,4 @@ has_output:
 - nmdc:dobj-11-ndgg7b37
 - nmdc:dobj-11-ndgg7b38
 type: nmdc:MetabolomicsAnalysis
+metabolomics_analysis_category: gc_ms_metabolomics

src/data/invalid/MetabolomicsAnalysis-invalid_id-4.yaml

@@ -12,3 +12,4 @@ has_output:
 - nmdc:dobj-11-ndgg7b37
 - nmdc:dobj-11-ndgg7b38
 type: nmdc:MetabolomicsAnalysis
+metabolomics_analysis_category: gc_ms_metabolomics

src/data/invalid/MetabolomicsAnalysis-invalid_no_informed_by.yaml

@@ -11,3 +11,4 @@ has_output:
 - nmdc:dobj-11-ndgg7b37
 - nmdc:dobj-11-ndgg7b38
 type: nmdc:MetabolomicsAnalysis
+metabolomics_analysis_category: gc_ms_metabolomics

src/data/invalid/MetabolomicsAnalysis-metab_quantified.yaml

@@ -1,3 +1,4 @@
+# this example is invalid because the has_metabolite_quantifications slot is used when it should be has_metabolite_identification
 id: nmdc:wfmb-99-ABCDEF.1
 type: nmdc:MetabolomicsAnalysis
 started_at_time: '2021-08-05T14:48:51+00:00'
@@ -18,3 +19,4 @@ has_metabolite_quantifications:
     highest_similarity_score: 0.9534156546099186
     metabolite_quantified: chebi:16997
     type: nmdc:MetaboliteQuantification
+metabolomics_analysis_category: gc_ms_metabolomics

src/data/valid/Database-Metabolomics-configuration.yaml

@@ -12,6 +12,7 @@ workflow_execution_set:
     has_output:
       - nmdc:dobj-90-izwYW61
     type: nmdc:MetabolomicsAnalysis
+    metabolomics_analysis_category: gc_ms_metabolomics
 data_object_set:
   - id: nmdc:dobj-70-izwYW6
     data_category: workflow_parameter_data

src/data/valid/Database-MetabolomicsAnalysis-1.yaml

@@ -13,3 +13,4 @@ workflow_execution_set:
       - nmdc:dobj-99-xxxxxx3
       - nmdc:dobj-99-xxxxxx4
     type: nmdc:MetabolomicsAnalysis
+    metabolomics_analysis_category: gc_ms_metabolomics

src/data/valid/Database-interleaved.yaml

@@ -3554,6 +3554,7 @@ workflow_execution_set:
     git_url: https://example.org/WorkflowExecutionActivity
     was_informed_by: nmdc:dgms-11-3u74ds
     started_at_time: '2021-08-05T14:48:51+00:00'
+    metabolomics_analysis_category: gc_ms_metabolomics
   - id: nmdc:wfmb-4d8-3z74d.1
     type: nmdc:MetabolomicsAnalysis
     name: soil metabolomics analysis
@@ -3570,6 +3571,7 @@ workflow_execution_set:
       - type: nmdc:MetaboliteIdentification
         highest_similarity_score: 0.88
         metabolite_identified: CHEBI:16236
+    metabolomics_analysis_category: gc_ms_metabolomics
   - id: nmdc:wfmgan-99-4d83z.1
     type: nmdc:MetagenomeAnnotation
     name: human gut metagenome annotation

src/data/valid/Database-lipid-workflow.yaml

@@ -16,6 +16,7 @@ workflow_execution_set:
     was_informed_by: "nmdc:dgms-11-vfzh1754"
     ended_at_time: "2024-12-19 03:09:58"
     version: "1.0.0"
+    metabolomics_analysis_category: lc_ms_lipidomics
 
 data_object_set:
   - id: "nmdc:dobj-11-r3xkmv70"

src/data/valid/Database-metabolomics_calibration_example.yaml

@@ -51,3 +51,4 @@ workflow_execution_set:
   git_url: https://github.com/microbiomedata/metaMS
   started_at_time: '2021-01-07T23:54:40Z'
   was_informed_by: nmdc:dgms-13-122e4240
+  metabolomics_analysis_category: gc_ms_metabolomics

src/data/valid/MetabolomicsAnalysis-1.yaml

@@ -18,3 +18,4 @@ has_metabolite_identifications:
     highest_similarity_score: 0.9534156546099186
     metabolite_identified: CHEBI:16997
     type: nmdc:MetaboliteIdentification
+metabolomics_analysis_category: gc_ms_metabolomics

src/data/valid/MetabolomicsAnalysis-2.yaml

@@ -12,3 +12,4 @@ has_output:
   - nmdc:dobj-11-ndgg7b38
 type: nmdc:MetabolomicsAnalysis
 was_informed_by: nmdc:dgms-11-dj832d
+metabolomics_analysis_category: gc_ms_metabolomics

src/schema/workflow_execution_activity.yaml

@@ -248,6 +248,7 @@ classes:
     slots:
       - has_metabolite_identifications
       - uses_calibration
+      - metabolomics_analysis_category
     slot_usage:
       id:
         required: true
@@ -299,6 +300,12 @@ slots:
       The category of metaproteomics analysis being performed.
     required: true
 
+  metabolomics_analysis_category:
+    range: MetabolomicsAnalysisCategoryEnum
+    description: >-
+      The category of metabolomics analysis being performed.
+    required: true
+
   metagenome_assembly_parameter:
     abstract: true
 
@@ -547,3 +554,16 @@ enums:
       in_silico_metagenome:
         description: >-
           A metaproteomics analysis that is matched to an in silico generated metagenome.
+  
+  MetabolomicsAnalysisCategoryEnum:
+    description: The category of metabolomics analysis being performed.
+    permissible_values:
+      gc_ms_metabolomics:
+        description: >-
+          A metabolomics analysis that is performed on gas chromatography mass spectrometry data.
+      lc_ms_lipidomics:
+        description: >-
+          A metabolomics analysis that is performed on liquid chromatography mass spectrometry data for lipidomics annotation.
+      lc_ms_metabolomics:
+        description: >-
+          A metabolomics analysis that is performed on liquid chromatography mass spectrometry data.