feat: added orcid and invesigator name now pid name (#807)

* feat: added orcid and invesigator name now pid name

* feat:fixed format and updated examples

---------

Co-authored-by: jtyoung84 <[email protected]>

Author: micahwoodard

examples/data_description.json

@@ -37,7 +37,12 @@
    "data_level": "raw",
    "group": null,
    "investigators": [
-      "Jane Smith"
+      {
+         "name": "Jane Smith",
+         "abbreviation": null,
+         "registry": null,
+         "registry_identifier": null
+      }
    ],
    "project_name": null,
    "restrictions": null,

examples/data_description.py

@@ -4,6 +4,7 @@
 from aind_data_schema.core.data_description import Funding, RawDataDescription
 from aind_data_schema.models.modalities import Modality
 from aind_data_schema.models.organizations import Organization
+from aind_data_schema.models.pid_names import PIDName
 from aind_data_schema.models.platforms import Platform
 
 d = RawDataDescription(
@@ -12,7 +13,7 @@
     subject_id="12345",
     creation_time=datetime(2022, 2, 21, 16, 30, 1),
     institution=Organization.AIND,
-    investigators=["Jane Smith"],
+    investigators=[PIDName(name="Jane Smith")],
     funding_source=[Funding(funder=Organization.AI)],
 )
 

src/aind_data_schema/core/data_description.py

@@ -10,6 +10,7 @@
 from aind_data_schema.base import AindCoreModel, AindModel
 from aind_data_schema.models.modalities import Modality
 from aind_data_schema.models.organizations import Organization
+from aind_data_schema.models.pid_names import PIDName
 from aind_data_schema.models.platforms import Platform
 
 
@@ -147,7 +148,7 @@ class DataDescription(AindCoreModel):
         description="A short name for the group of individuals that collected this data",
         title="Group",
     )
-    investigators: List[str] = Field(
+    investigators: List[PIDName] = Field(
         ...,
         description="Full name(s) of key investigators (e.g. PI, lead scientist, contact person)",
         title="Investigators",

src/aind_data_schema/models/registry.py

@@ -48,13 +48,21 @@ class NationalCenterForBiotechnologyInformation(_Registry):
     abbreviation: Literal["NCBI"] = "NCBI"
 
 
+class OpenResearcherAndContributorID(_Registry):
+    """NationalCenterForBiotechnologyInformation"""
+
+    name: Literal["Open Researcher and Contributor ID"] = "Open Researcher and Contributor ID"
+    abbreviation: Literal["ORCID"] = "ORCID"
+
+
 class Registry:
     """Registry definitions"""
 
     ADDGENE = Addgene()
     ROR = ResearchOrganizationRegistry()
     MGI = MouseGenomeInformatics()
     NCBI = NationalCenterForBiotechnologyInformation()
+    ORCID = OpenResearcherAndContributorID()
     RRID = ResearchResourceIdentifiers()
 
     _ALL = tuple(_Registry.__subclasses__())

tests/test_data_description.py

@@ -22,6 +22,7 @@
 )
 from aind_data_schema.models.modalities import Modality
 from aind_data_schema.models.organizations import Organization
+from aind_data_schema.models.pid_names import PIDName
 from aind_data_schema.models.platforms import Platform
 
 DATA_DESCRIPTION_FILES_PATH = Path(__file__).parent / "resources" / "ephys_data_description"
@@ -60,7 +61,7 @@ def test_constructors(self):
             modality=[Modality.ECEPHYS],
             platform=Platform.ECEPHYS,
             subject_id="12345",
-            investigators=["Jane Smith"],
+            investigators=[PIDName(name="Jane Smith")],
         )
 
         r1 = DerivedDataDescription(
@@ -72,7 +73,7 @@ def test_constructors(self):
             modality=da.modality,
             platform=da.platform,
             subject_id=da.subject_id,
-            investigators=["Jane Smith"],
+            investigators=[PIDName(name="Jane Smith")],
         )
 
         r2 = DerivedDataDescription(
@@ -84,7 +85,7 @@ def test_constructors(self):
             modality=r1.modality,
             platform=r1.platform,
             subject_id="12345",
-            investigators=["Jane Smith"],
+            investigators=[PIDName(name="Jane Smith")],
         )
 
         r3 = DerivedDataDescription(
@@ -96,7 +97,7 @@ def test_constructors(self):
             modality=r2.modality,
             platform=r2.platform,
             subject_id="12345",
-            investigators=["Jane Smith"],
+            investigators=[PIDName(name="Jane Smith")],
         )
         assert r3 is not None
 
@@ -109,7 +110,7 @@ def test_constructors(self):
             creation_time=dt,
             institution=Organization.AIND,
             funding_source=[f],
-            investigators=["Jane Smith"],
+            investigators=[PIDName(name="Jane Smith")],
         )
 
         assert dd is not None
@@ -125,7 +126,7 @@ def test_constructors(self):
                 creation_time=dt,
                 institution=Organization.AIND,
                 funding_source=[f],
-                investigators=["Jane Smith"],
+                investigators=[PIDName(name="Jane Smith")],
             )
 
         ad = AnalysisDescription(
@@ -137,7 +138,7 @@ def test_constructors(self):
             platform=Platform.EXASPIM,
             institution=Organization.AIND,
             funding_source=[f],
-            investigators=["Jane Smith"],
+            investigators=[PIDName(name="Jane Smith")],
         )
         self.assertEqual(ad.name, build_data_name("project_analysis", dt))
 
@@ -151,7 +152,7 @@ def test_constructors(self):
                 creation_time=dt,
                 institution=Organization.AIND,
                 funding_source=[f],
-                investigators=["Jane Smith"],
+                investigators=[PIDName(name="Jane Smith")],
             )
 
         with self.assertRaises(ValueError):
@@ -185,7 +186,7 @@ def test_constructors(self):
                 creation_time=dt,
                 institution=Organization.AIND,
                 funding_source=[f],
-                investigators=["Jane Smith"],
+                investigators=[PIDName(name="Jane Smith")],
             )
 
         with self.assertRaises(ValueError):
@@ -198,7 +199,7 @@ def test_constructors(self):
                 creation_time=dt,
                 institution=Organization.AIND,
                 funding_source=[f],
-                investigators=["Jane Smith"],
+                investigators=[PIDName(name="Jane Smith")],
             )
 
     def test_pattern_errors(self):
@@ -214,7 +215,7 @@ def test_pattern_errors(self):
                 creation_time=datetime.datetime(2020, 10, 10, 10, 10, 10),
                 institution=Organization.AIND,
                 funding_source=[Funding(funder=Organization.NINDS, grant_number="grant001")],
-                investigators=["Jane Smith"],
+                investigators=[PIDName(name="Jane Smith")],
             )
         expected_exception = (
             "1 validation error for DataDescription\n"
@@ -278,7 +279,7 @@ def test_name_label_error(self):
                 creation_time=datetime.datetime(2020, 10, 10, 10, 10, 10),
                 institution=Organization.AIND,
                 funding_source=[Funding(funder=Organization.NINDS, grant_number="grant001")],
-                investigators=["Jane Smith"],
+                investigators=[PIDName(name="Jane Smith")],
             )
         self.assertTrue("Value error, Either label or name must be set" in repr(e.exception))
 
@@ -295,7 +296,7 @@ def test_round_trip(self):
             modality=[Modality.SPIM],
             platform=Platform.EXASPIM,
             subject_id="12345",
-            investigators=["Jane Smith"],
+            investigators=[PIDName(name="Jane Smith")],
         )
 
         da2 = RawDataDescription.model_validate_json(da1.model_dump_json())
@@ -355,7 +356,7 @@ def test_from_data_description(self):
             creation_time=datetime.datetime(2020, 10, 10, 10, 10, 10),
             institution=Organization.AIND,
             funding_source=[Funding(funder=Organization.NINDS, grant_number="grant001")],
-            investigators=["Jane Smith"],
+            investigators=[PIDName(name="Jane Smith")],
         )
 
         process_name = "spikesorter"
@@ -377,7 +378,7 @@ def test_derived_data_description_build_name(self):
             modality=[Modality.ECEPHYS],
             platform=Platform.ECEPHYS,
             subject_id="12345",
-            investigators=["Jane Smith"],
+            investigators=[PIDName(name="Jane Smith")],
         )
         self.assertEqual("input_2020-10-10_10-10-10", dd.name)