Draft of Gene Ontology Controlled Keyword. Waiting for the final GO list.

The failed test is from test_hgvs_genes.py. Mypy errors are all from statistics.py.

Author: EstelleDa

alembic/versions/18b6d81c045e_controlled_keyword_gene_ontology.py

@@ -0,0 +1,33 @@
+"""Controlled keyword Gene Ontology
+
+Revision ID: 18b6d81c045e
+Revises: c404b6719110
+Create Date: 2025-03-14 19:13:01.554046
+
+"""
+from alembic import op
+import sqlalchemy as sa
+
+
+# revision identifiers, used by Alembic.
+revision = '18b6d81c045e'
+down_revision = 'c404b6719110'
+branch_labels = None
+depends_on = None
+
+
+def upgrade():
+    # ### commands auto generated by Alembic - please adjust! ###
+    op.add_column('controlled_keywords', sa.Column('accession', sa.String(), nullable=True))
+
+    # TODO: Will modify this part when we get the final GO terms.
+    op.execute(
+        """INSERT INTO controlled_keywords (key, value, vocabulary, accession, special, description, creation_date, modification_date) VALUES ('Phenotypic Assay Mechanism', 'Other', NULL, NULL, False, 'The Gene Ontology (GO) is a structured, standardized representation of biological knowledge.', NOW(), NOW())"""
+    )
+    # ### end Alembic commands ###
+
+
+def downgrade():
+    # ### commands auto generated by Alembic - please adjust! ###
+    op.drop_column('controlled_keywords', 'accession')
+    # ### end Alembic commands ###

src/mavedb/lib/validation/keywords.py

@@ -1,9 +1,18 @@
+import re
 from typing import Optional
 
 from mavedb.lib.validation.exceptions import ValidationError
 from mavedb.lib.validation.utilities import is_null
 
 
+def validate_accession(key: str, value: str, accession: Optional[str]):
+    if key.lower() == "phenotypic assay mechanism" and value.lower() != "other":
+        # The Gene Ontology accession is a unique seven digit identifier prefixed by GO:.
+        # e.g. GO:0005739, GO:1904659, or GO:0016597.
+        if accession is None or not re.match(r"^GO:\d{7}$", accession):
+            raise ValidationError("Invalid Gene Ontology accession.")
+
+
 # TODO: value will not be Optional when we confirm the final controlled keyword list.
 def validate_description(value: str, key: str, description: Optional[str]):
     if value.lower() == "other" and (description is None or description.strip() == ""):

src/mavedb/models/controlled_keyword.py

@@ -12,6 +12,7 @@ class ControlledKeyword(Base):
     key = Column(String, nullable=False)
     value = Column(String, nullable=False)
     vocabulary = Column(String, nullable=True)
+    accession = Column(String, nullable=True)
     special = Column(Boolean, nullable=True)
     description = Column(String, nullable=True)
     creation_date = Column(Date, nullable=False, default=date.today)

src/mavedb/models/experiment.py

@@ -163,6 +163,7 @@ def keywords(self) -> list[dict]:
                         "key": controlled_keyword.key,
                         "value": controlled_keyword.value,
                         "vocabulary": controlled_keyword.vocabulary,
+                        "accession": controlled_keyword.accession,
                         "special": controlled_keyword.special,
                         "description": controlled_keyword.description,
                     },

src/mavedb/view_models/keyword.py

@@ -17,6 +17,7 @@ class KeywordBase(BaseModel):
     key: str
     value: Optional[str]
     vocabulary: Optional[str]
+    accession: Optional[str]
     special: Optional[bool]
     description: Optional[str]
 
@@ -25,6 +26,13 @@ def validate_key(cls, v):
         keywords.validate_keyword(v)
         return v
 
+    @validator("accession")
+    def validate_accession(cls, accession, values):
+        key = values.get("key")
+        value = values.get("value")
+        keywords.validate_accession(key, value, accession)
+        return accession
+
     # validator("value") blocks creating a new experiment without controlled keywords so comment it first.
     # @validator("value")
     # def validate_value(cls, v):

tests/helpers/constants.py

@@ -162,6 +162,8 @@
         "description": "Description",
     },
     {"key": "Delivery method", "value": "Other", "special": False, "description": "Description"},
+    {"key": "Phenotypic Assay Mechanism", "value": "Other", "accession": None, "special": False, "description": "Description"},
+    {"key": "Phenotypic Assay Mechanism", "value": "Value", "accession": "GO:1234567", "special": False, "description": "Description"},
 ]
 
 TEST_KEYWORDS = [

tests/routers/test_experiments.py

@@ -270,6 +270,70 @@ def test_cannot_create_experiment_that_keywords_has_wrong_combination4(client, s
     )
 
 
+def test_create_experiment_that_keyword_gene_ontology_has_valid_accession(client, setup_router_db):
+    valid_keyword = {
+        "keywords": [
+            {
+                "keyword": {
+                    "key": "Phenotypic Assay Mechanism",
+                    "value": "Value",
+                    "accession": "GO:1234567",
+                    "special": False,
+                    "description": "Description"},
+            },
+        ],
+    }
+    experiment = {**TEST_MINIMAL_EXPERIMENT, **valid_keyword}
+    response = client.post("/api/v1/experiments/", json=experiment)
+    assert response.status_code == 200
+    response_data = response.json()
+    assert response_data["keywords"][0]["keyword"]["key"] == "Phenotypic Assay Mechanism"
+    assert response_data["keywords"][0]["keyword"]["value"] == "Value"
+    assert response_data["keywords"][0]["keyword"]["accession"] == "GO:1234567"
+
+
+def test_create_experiment_that_keyword_gene_ontology_is_other_without_accession(client, setup_router_db):
+    valid_keyword = {
+        "keywords": [
+            {
+                "keyword": {
+                    "key": "Phenotypic Assay Mechanism",
+                    "value": "Other",
+                    "accession": None,
+                    "description": "Description"
+                },
+                "description": "Description",
+            },
+        ]
+    }
+    experiment = {**TEST_MINIMAL_EXPERIMENT, **valid_keyword}
+    response = client.post("/api/v1/experiments/", json=experiment)
+    assert response.status_code == 200
+    response_data = response.json()
+    assert response_data["keywords"][0]["keyword"]["key"] == "Phenotypic Assay Mechanism"
+    assert response_data["keywords"][0]["keyword"]["value"] == "Other"
+
+
+def test_cannot_create_experiment_that_keyword_has_an_invalid_accession(client, setup_router_db):
+    invalid_keyword = {
+        "keywords": [
+            {
+                "keyword": {
+                    "key": "Phenotypic Assay Mechanism",
+                    "value": "Value",
+                    "accession": "invalid",
+                    "description": "Description"
+                },
+            },
+        ]
+    }
+    experiment = {**TEST_MINIMAL_EXPERIMENT, **invalid_keyword}
+    response = client.post("/api/v1/experiments/", json=experiment)
+    assert response.status_code == 422
+    response_data = response.json()
+    assert response_data["detail"][0]["msg"] == "Invalid Gene Ontology accession."
+
+
 def test_cannot_create_experiment_that_keyword_value_is_other_without_description(client, setup_router_db):
     """
     Test src/mavedb/lib/validation/keywords.validate_description function

tests/validation/test_keywords.py

@@ -2,6 +2,7 @@
 
 from mavedb.lib.validation.exceptions import ValidationError
 from mavedb.lib.validation.keywords import (
+    validate_accession,
     validate_description,
     validate_duplicates,
     validate_keyword,
@@ -39,6 +40,25 @@ def test_other_without_extra_description(self):
         with self.assertRaises(ValidationError):
             validate_description(value, key, description)
 
+    def test_Gene_Ontology_valid_accession(self):
+        key = "Phenotypic Assay Mechanism"
+        value = "value"
+        accession = "GO:1234567"
+        validate_accession(key, value, accession)
+
+    def test_Gene_Ontology_invalid_accession(self):
+        key = "Phenotypic Assay Mechanism"
+        value = "value"
+        accession = "GO:123"
+        with self.assertRaises(ValidationError):
+            validate_accession(key, value, accession)
+
+    def test_Gene_Ontoloty_term_is_other(self):
+        key = "Phenotypic Assay Mechanism"
+        value = "Other"
+        accession = None
+        validate_accession(key, value, accession)
+
     def test_duplicate_keys(self):
         # Invalid keywords list.
         keyword1 = {