Merge pull request #1254 from dandi/enh/metadata-cellline

enh: add cellline parsing
dandi · Apr 3, 2023 · 84ed322 · 84ed322
2 parents 47e9d97 + 65eda7c
commit 84ed322
Show file tree

Hide file tree

Showing 3 changed files with 162 additions and 1 deletion.
diff --git a/dandi/metadata.py b/dandi/metadata.py
@@ -454,12 +454,23 @@ def extract_sex(metadata: dict) -> Optional[models.SexType]:
 
 def extract_strain(metadata: dict) -> Optional[models.StrainType]:
     value = metadata.get("strain", None)
-    if value is not None and value != "":
+    if value:
+        # Don't assign cell lines to strain
+        if value.lower().startswith("cellline:"):
+            return None
         return models.StrainType(name=value)
     else:
         return None
 
 
+def extract_cellLine(metadata: dict) -> Optional[str]:
+    value: str = metadata.get("strain", "")
+    if value and value.lower().startswith("cellline:"):
+        return value.split(":", 1)[1].strip()
+    else:
+        return None
+
+
 species_map = [
     (
         ["mouse"],
@@ -709,6 +720,7 @@ def extract_digest(metadata: dict) -> Optional[Dict[models.DigestType, str]]:
     "age": extract_age,
     "sex": extract_sex,
     "strain": extract_strain,
+    "cellLine": extract_cellLine,
     "assayType": extract_assay_type,
     "anatomy": extract_anatomy,
     "digest": extract_digest,

diff --git a/dandi/tests/data/metadata/metadata2asset_cellline.json b/dandi/tests/data/metadata/metadata2asset_cellline.json
@@ -0,0 +1,93 @@
+{
+  "id": "dandiasset:0b0a1a0b-e3ea-4cf6-be94-e02c830d54be",
+  "schemaKey": "Asset",
+  "schemaVersion": "0.4.1",
+  "keywords": [
+    "test",
+    "sample",
+    "example",
+    "test-case"
+  ],
+  "access": [
+    {
+      "schemaKey": "AccessRequirements",
+      "status": "dandi:OpenAccess"
+    }
+  ],
+  "wasGeneratedBy": [
+    {
+      "schemaKey": "Session",
+      "identifier": "XYZ789",
+      "name": "XYZ789",
+      "description": "Some test data",
+      "startDate": "2020-08-31T15:58:28-04:00",
+      "used": [
+        {
+          "schemaKey": "Equipment",
+          "identifier": "probe:probe04",
+          "name": "Ecephys Probe"
+        }
+      ]
+    }
+  ],
+  "contentSize": 69105,
+  "encodingFormat": "application/x-nwb",
+  "digest": {
+    "dandi:dandi-etag": "e455839e5ab2fa659861f58a423fd17f-1"
+  },
+  "path": "/test/path",
+  "wasDerivedFrom": [
+    {
+      "schemaKey": "BioSample",
+      "identifier": "cell01",
+      "sampleType": {
+        "schemaKey": "SampleType",
+        "name": "cell"
+      },
+      "wasDerivedFrom": [
+        {
+          "schemaKey": "BioSample",
+          "identifier": "slice02",
+          "sampleType": {
+            "schemaKey": "SampleType",
+            "name": "slice"
+          },
+          "wasDerivedFrom": [
+            {
+              "schemaKey": "BioSample",
+              "identifier": "tissue03",
+              "sampleType": {
+                "schemaKey": "SampleType",
+                "name": "tissuesample"
+              }
+            }
+          ]
+        }
+      ]
+    }
+  ],
+  "wasAttributedTo": [
+    {
+      "schemaKey": "Participant",
+      "identifier": "a1b2c3",
+      "age": {
+        "unitText": "ISO-8601 duration",
+        "value": "P170DT12212S",
+        "schemaKey": "PropertyValue",
+        "valueReference": {"schemaKey": "PropertyValue", "value": "dandi:BirthReference"}
+      },
+      "sex": {
+        "schemaKey": "SexType",
+        "identifier": "http://purl.obolibrary.org/obo/PATO_0000384",
+        "name": "Male"
+      },
+      "cellLine": "abcdef/1",
+      "genotype": "Typical",
+      "species": {
+        "schemaKey": "SpeciesType",
+        "identifier": "http://purl.obolibrary.org/obo/NCBITaxon_1234175",
+        "name": "Cyperus bulbosus"
+      }
+    }
+  ]
+}
diff --git a/dandi/tests/test_metadata.py b/dandi/tests/test_metadata.py
@@ -34,6 +34,7 @@
 from ..dandiapi import RemoteBlobAsset
 from ..metadata import (
     extract_age,
+    extract_cellLine,
     extract_species,
     get_metadata,
     nwb2asset,
@@ -156,6 +157,18 @@ def test_parse_age(age: str, duration: Union[str, Tuple[str, str]]) -> None:
     assert parse_age(age) == (duration, ref)
 
 
+@pytest.mark.parametrize(
+    "s, t",
+    [
+        ("cellline: abcdef/1", "abcdef/1"),
+        ("CellLine:  cellline:1 ", "cellline:1"),
+        ("cell line: 1", None),
+    ],
+)
+def test_extract_cellLine(s, t):
+    assert extract_cellLine({"strain": s}) == t
+
+
 @pytest.mark.parametrize(
     "age, errmsg",
     [
@@ -336,6 +349,49 @@ def test_timedelta2duration(td: timedelta, duration: str) -> None:
             },
             marks=pytest.mark.obolibrary,
         ),
+        # add cellline exception to a new json object
+        pytest.param(
+            "metadata2asset_cellline.json",
+            {
+                "contentSize": 69105,
+                "digest": "e455839e5ab2fa659861f58a423fd17f-1",
+                "digest_type": "dandi_etag",
+                "encodingFormat": "application/x-nwb",
+                "experiment_description": "Experiment Description",
+                "experimenter": "Joe Q. Experimenter",
+                "id": "dandiasset:0b0a1a0b-e3ea-4cf6-be94-e02c830d54be",
+                "institution": "University College",
+                "keywords": ["test", "sample", "example", "test-case"],
+                "lab": "Retriever Laboratory",
+                "related_publications": "A Brief History of Test Cases",
+                "session_description": "Some test data",
+                "session_id": "XYZ789",
+                "session_start_time": "2020-08-31T15:58:28-04:00",
+                "age": "23 days",
+                "date_of_birth": "2020-03-14T12:34:56-04:00",
+                "genotype": "Typical",
+                "sex": "M",
+                "strain": "cellline: abcdef/1",
+                "species": "http://purl.obolibrary.org/obo/NCBITaxon_1234175",  # Corner case
+                "subject_id": "a1b2c3",
+                "cell_id": "cell01",
+                "slice_id": "slice02",
+                "tissue_sample_id": "tissue03",
+                "probe_ids": "probe04",
+                "number_of_electrodes": 42,
+                "number_of_units": 6,
+                "nwb_version": "2.2.5",
+                "nd_types": [
+                    "Device (2)",
+                    "DynamicTable",
+                    "ElectricalSeries",
+                    "ElectrodeGroup",
+                    "Subject",
+                ],
+                "path": "/test/path",
+            },
+            marks=pytest.mark.obolibrary,
+        ),
     ],
 )
 def test_prepare_metadata(filename: str, metadata: Dict[str, Any]) -> None: