Added new sample metadata.

Added new sample metadata, sample_uuid and sample_lims. Provided mapping of the values of these metadata to mlwh sample table columns. Updated test data.
wtsi-npg · Dec 14, 2024 · e01509f · e01509f
1 parent df38af5
commit e01509f
Show file tree

Hide file tree

Showing 10 changed files with 28 additions and 2 deletions.
diff --git a/src/npg_irods/db/mlwh.py b/src/npg_irods/db/mlwh.py
@@ -80,6 +80,7 @@ class Sample(Base):
     donor_id = mapped_column(String(255))
     date_of_consent_withdrawn = mapped_column(DateTime)
     marked_as_consent_withdrawn_by = mapped_column(String(255))
+    uuid_sample_lims = mapped_column(String(36), nullable=False)
 
     iseq_flowcell: Mapped["IseqFlowcell"] = relationship(
         "IseqFlowcell", back_populates="sample"

diff --git a/src/npg_irods/metadata/lims.py b/src/npg_irods/metadata/lims.py
@@ -62,9 +62,11 @@ class TrackedSample(AsValueEnum):
     CONTROL = "sample_control"
     DONOR_ID = "sample_donor_id"
     ID = "sample_id"
+    LIMS = "sample_lims"
     NAME = "sample"
     PUBLIC_NAME = "sample_public_name"
     SUPPLIER_NAME = "sample_supplier_name"
+    UUID = "sample_uuid"
 
 
 @unique
@@ -88,6 +90,8 @@ def make_sample_metadata(sample: Sample) -> list[AVU]:
      - sample name
      - sample public name
      - sample supplier name
+     - sample uuid
+     - ID of the LIM system where this sample was registered
 
     Args:
         sample: An ML warehouse schema Sample.
@@ -104,9 +108,11 @@ def make_sample_metadata(sample: Sample) -> list[AVU]:
         ],
         [TrackedSample.DONOR_ID, sample.donor_id],
         [TrackedSample.ID, sample.id_sample_lims],
+        [TrackedSample.LIMS, sample.id_lims],
         [TrackedSample.NAME, sample.name],
         [TrackedSample.PUBLIC_NAME, sample.public_name],
         [TrackedSample.SUPPLIER_NAME, sample.supplier_name],
+        [TrackedSample.UUID, sample.uuid_sample_lims],
     ]
 
     return [avu for avu in starmap(avu_if_value, av) if avu is not None]

diff --git a/tests/conftest.py b/tests/conftest.py
@@ -157,6 +157,7 @@ def initialize_mlwh_study_and_sample(session: Session):
         name="name1",
         public_name="public_name1",
         supplier_name="supplier_name1",
+        uuid_sample_lims="82429892-0ab6-11ee-b5ba-fa163eac3af7",
         **default_timestamps,
     )
     session.add(sample_y)

diff --git a/tests/illumina/conftest.py b/tests/illumina/conftest.py
@@ -72,12 +72,17 @@ def make_sample(n):
             name=f"name{n}",
             public_name=f"public_name{n}",
             supplier_name=f"supplier_name{n}",
+            uuid_sample_lims=f"52429892-0ab6-11ee-b5ba-fa163eac3af{n}",
             **default_timestamps,
         )
 
     samples = [make_sample(n) for n in range(1, 4)]
     control_sample = Sample(
-        id_lims="LIMS_888", id_sample_lims="phix", name="Phi X", **default_timestamps
+        id_lims="LIMS_888",
+        id_sample_lims="phix",
+        name="Phi X",
+        uuid_sample_lims="42429892-0ab6-11ee-b5ba-fa163eac3af3",
+        **default_timestamps,
     )
     session.add_all([*samples, control_sample])
 

diff --git a/tests/illumina/test_metadata_update.py b/tests/illumina/test_metadata_update.py
@@ -187,9 +187,11 @@ def test_updates_present_metadata(
             AVU(TrackedSample.COMMON_NAME, "common_name1"),
             AVU(TrackedSample.DONOR_ID, "donor_id1"),
             AVU(TrackedSample.ID, "id_sample_lims1"),
+            AVU(TrackedSample.LIMS, "LIMS_01"),
             AVU(TrackedSample.NAME, "name1"),
             AVU(TrackedSample.PUBLIC_NAME, "public_name1"),
             AVU(TrackedSample.SUPPLIER_NAME, "supplier_name1"),
+            AVU(TrackedSample.UUID, "52429892-0ab6-11ee-b5ba-fa163eac3af1"),
             AVU(TrackedStudy.ID, "4000"),
             AVU(TrackedStudy.NAME, "Study A"),
         ]
@@ -469,9 +471,11 @@ def test_updates_present_study_permissions(
             AVU(TrackedSample.COMMON_NAME, "common_name1"),
             AVU(TrackedSample.DONOR_ID, "donor_id1"),
             AVU(TrackedSample.ID, "id_sample_lims1"),
+            AVU(TrackedSample.LIMS, "LIMS_01"),
             AVU(TrackedSample.NAME, "name1"),
             AVU(TrackedSample.PUBLIC_NAME, "public_name1"),
             AVU(TrackedSample.SUPPLIER_NAME, "supplier_name1"),
+            AVU(TrackedSample.UUID, "52429892-0ab6-11ee-b5ba-fa163eac3af1"),
             AVU(TrackedStudy.ID, "4000"),
             AVU(TrackedStudy.NAME, "Study A"),
         ]

diff --git a/tests/ont/conftest.py b/tests/ont/conftest.py
@@ -86,6 +86,7 @@ def make_sample(n):
             name=f"name{n}",
             public_name=f"public_name{n}",
             supplier_name=f"supplier_name{n}",
+            uuid_sample_lims=f"62429892-0ab6-11ee-b5ba-fa163eac3af{n}",
             **default_timestamps,
         )
 

diff --git a/tests/pacbio/conftest.py b/tests/pacbio/conftest.py
@@ -84,6 +84,7 @@ def make_sample(n):
             name=f"name{n}",
             public_name=f"public_name{n}",
             supplier_name=f"supplier_name{n}",
+            uuid_sample_lims=f"72429892-0ab6-11ee-b5ba-fa163eac3af{n}",
             **default_timestamps,
         )
 

diff --git a/tests/pacbio/test_metadata_update.py b/tests/pacbio/test_metadata_update.py
@@ -136,9 +136,11 @@ def test_updates_present_metadata(
             AVU(TrackedSample.COMMON_NAME, "common_name1"),
             AVU(TrackedSample.DONOR_ID, "donor_id1"),
             AVU(TrackedSample.ID, "id_sample_lims1"),
+            AVU(TrackedSample.LIMS, "LIMS_01"),
             AVU(TrackedSample.NAME, "name1"),
             AVU(TrackedSample.PUBLIC_NAME, "public_name1"),
             AVU(TrackedSample.SUPPLIER_NAME, "supplier_name1"),
+            AVU(TrackedSample.UUID, "72429892-0ab6-11ee-b5ba-fa163eac3af1"),
             AVU(TrackedStudy.ID, "1000"),
             AVU(TrackedStudy.NAME, "Study X"),
         ]
@@ -245,9 +247,11 @@ def test_updates_present_study_permissions(
             AVU(TrackedSample.COMMON_NAME, "common_name1"),
             AVU(TrackedSample.DONOR_ID, "donor_id1"),
             AVU(TrackedSample.ID, "id_sample_lims1"),
+            AVU(TrackedSample.LIMS, "LIMS_01"),
             AVU(TrackedSample.NAME, "name1"),
             AVU(TrackedSample.PUBLIC_NAME, "public_name1"),
             AVU(TrackedSample.SUPPLIER_NAME, "supplier_name1"),
+            AVU(TrackedSample.UUID, "72429892-0ab6-11ee-b5ba-fa163eac3af1"),
             AVU(TrackedStudy.ID, "1000"),
             AVU(TrackedStudy.NAME, "Study X"),
         ]

diff --git a/tests/test_common.py b/tests/test_common.py
@@ -38,17 +38,18 @@ def test_ensure_secondary_metadata_updated(
             AVU(TrackedStudy.TITLE, "Test Study Title"),
             AVU(TrackedStudy.ACCESSION_NUMBER, "Test Accession"),
             AVU(TrackedSample.ID, sample_id),
+            AVU(TrackedSample.LIMS, "LIMS_01"),
             AVU(TrackedSample.ACCESSION_NUMBER, "Test Accession"),
             AVU(TrackedSample.COMMON_NAME, "common_name1"),
             AVU(TrackedSample.DONOR_ID, "donor_id1"),
             AVU(TrackedSample.NAME, "name1"),
             AVU(TrackedSample.PUBLIC_NAME, "public_name1"),
             AVU(TrackedSample.SUPPLIER_NAME, "supplier_name1"),
+            AVU(TrackedSample.UUID, "82429892-0ab6-11ee-b5ba-fa163eac3af7"),
         ]
 
         obj = DataObject(simple_study_and_sample_data_object)
         assert ensure_secondary_metadata_updated(obj, simple_study_and_sample_mlwh)
-
         for avu in expected_avus:
             assert avu in obj.metadata()
 

diff --git a/tests/test_utilities.py b/tests/test_utilities.py
@@ -668,12 +668,14 @@ def test_update_secondary_metadata(
             AVU(TrackedStudy.TITLE, "Test Study Title"),
             AVU(TrackedStudy.ACCESSION_NUMBER, "Test Accession"),
             AVU(TrackedSample.ID, "id_sample_lims1"),
+            AVU(TrackedSample.LIMS, "LIMS_01"),
             AVU(TrackedSample.ACCESSION_NUMBER, "Test Accession"),
             AVU(TrackedSample.COMMON_NAME, "common_name1"),
             AVU(TrackedSample.DONOR_ID, "donor_id1"),
             AVU(TrackedSample.NAME, "name1"),
             AVU(TrackedSample.PUBLIC_NAME, "public_name1"),
             AVU(TrackedSample.SUPPLIER_NAME, "supplier_name1"),
+            AVU(TrackedSample.UUID, "82429892-0ab6-11ee-b5ba-fa163eac3af7"),
         ]
 
         obj = DataObject(simple_study_and_sample_data_object)