nci-cidc-api-modules 1.2.29__py3-none-any.whl → 1.2.34__py3-none-any.whl

cidc_api/models/data.py

@@ -1,4 +1,5 @@
 from cidc_api.models.pydantic.stage2 import all_models
+from cidc_api.models.db.stage2 import all_models as all_db_models
 
 standard_data_categories = [model.__data_category__ for model in all_models if hasattr(model, "__data_category__")]
 
@@ -13,3 +14,6 @@ class Dataset(dict):
 
 # Maps data categories like "treatment" to their associated pydantic model
 data_category_to_model = {model.__data_category__: model for model in all_models if hasattr(model, "__data_category__")}
+data_category_to_db_model = {
+    model.__data_category__: model for model in all_db_models if hasattr(model, "__data_category__")
+}

cidc_api/models/db/stage2/additional_treatment_orm.py

@@ -11,12 +11,13 @@ class AdditionalTreatmentORM(BaseORM):
     __tablename__ = "additional_treatment"
     __repr_attrs__ = ["additional_treatment_id", "participant_id", "description"]
     __table_args__ = {"schema": "stage2"}
+    __data_category__ = "additional_treatment"
 
     additional_treatment_id: Mapped[int] = mapped_column(primary_key=True)
     participant_id: Mapped[int] = mapped_column(ForeignKey("stage2.participant.participant_id", ondelete="CASCADE"))
 
-    days_to_start: Mapped[Optional[NonNegativeInt]]
-    days_to_end: Mapped[Optional[NonNegativeInt]]
-    description: Mapped[str]
+    additional_treatment_days_to_start: Mapped[Optional[NonNegativeInt]]
+    additional_treatment_days_to_end: Mapped[Optional[NonNegativeInt]]
+    additional_treatment_description: Mapped[str]
 
     participant: Mapped["ParticipantORM"] = relationship(back_populates="additional_treatments", cascade="all, delete")

cidc_api/models/db/stage2/adverse_event_orm.py

@@ -22,6 +22,7 @@ class AdverseEventORM(BaseORM):
     __tablename__ = "adverse_event"
     __repr_attrs__ = ["adverse_event_id", "participant_id", "event_term"]
     __table_args__ = {"schema": "stage2"}
+    __data_category__ = "adverse_event"
 
     adverse_event_id: Mapped[int] = mapped_column(primary_key=True)
     participant_id: Mapped[int] = mapped_column(ForeignKey("stage2.participant.participant_id", ondelete="CASCADE"))

cidc_api/models/db/stage2/baseline_clinical_assessment_orm.py

@@ -11,6 +11,7 @@ class BaselineClinicalAssessmentORM(BaseORM):
     __tablename__ = "baseline_clinical_assessment"
     __repr_attrs__ = ["baseline_clinical_assessment_id", "participant_id"]
     __table_args__ = {"schema": "stage2"}
+    __data_category__ = "baseline_clinical_assessment"
 
     baseline_clinical_assessment_id: Mapped[int] = mapped_column(primary_key=True)
     participant_id: Mapped[int] = mapped_column(ForeignKey("stage2.participant.participant_id", ondelete="CASCADE"))

cidc_api/models/db/stage2/comorbidity_orm.py

@@ -11,6 +11,7 @@ class ComorbidityORM(BaseORM):
     __tablename__ = "comorbidity"
     __repr_attrs__ = ["comorbidity_id", "comorbidity_term"]
     __table_args__ = {"schema": "stage2"}
+    __data_category__ = "comorbidity"
 
     comorbidity_id: Mapped[int] = mapped_column(primary_key=True)
     medical_history_id: Mapped[int] = mapped_column(

cidc_api/models/db/stage2/consent_group_orm.py

@@ -16,6 +16,7 @@ class ConsentGroupORM(BaseORM):
         ),
         {"schema": "stage2"},
     )
+    __data_category__ = "consent_group"
 
     consent_group_id: Mapped[int] = mapped_column(primary_key=True)
     trial_id: Mapped[str]

cidc_api/models/db/stage2/demographic_orm.py

@@ -1,7 +1,9 @@
 from pydantic import NonNegativeFloat, PositiveFloat, PositiveInt
+from typing import List
 
 from sqlalchemy import ForeignKey
 from sqlalchemy.orm import Mapped, mapped_column, relationship
+from sqlalchemy.types import JSON
 
 from cidc_api.models.db.base_orm import BaseORM
 from cidc_api.models.types import (
@@ -21,22 +23,23 @@ class DemographicORM(BaseORM):
     __tablename__ = "demographic"
     __repr_attrs__ = ["demographic_id", "participant_id", "age_at_enrollment", "sex"]
     __table_args__ = {"schema": "stage2"}
+    __data_category__ = "demographic"
 
     demographic_id: Mapped[int] = mapped_column(primary_key=True)
     participant_id: Mapped[int] = mapped_column(ForeignKey("stage2.participant.participant_id", ondelete="CASCADE"))
     age_at_enrollment: Mapped[PositiveInt | None]
     age_at_enrollment_units: Mapped[AgeAtEnrollmentUnits | None]
-    age_90_or_older: Mapped[bool]
+    age_90_or_over: Mapped[bool]
     sex: Mapped[Sex]
-    race: Mapped[Race]
+    race: Mapped[List[Race]] = mapped_column(JSON)
     ethnicity: Mapped[Ethnicity]
     height: Mapped[PositiveFloat]
     height_units: Mapped[HeightUnits]
     weight: Mapped[PositiveFloat]
     weight_units: Mapped[WeightUnits]
-    body_mass_index: Mapped[PositiveFloat]
-    body_surface_area: Mapped[PositiveFloat]
-    body_surface_area_units: Mapped[BodySurfaceAreaUnits]
+    body_mass_index: Mapped[PositiveFloat | None]
+    body_surface_area: Mapped[PositiveFloat | None]
+    body_surface_area_units: Mapped[BodySurfaceAreaUnits | None]
     occupation: Mapped[Occupation | None]
     income: Mapped[NonNegativeFloat | None]
     highest_level_of_education: Mapped[Education | None]

cidc_api/models/db/stage2/disease_orm.py

@@ -7,7 +7,6 @@ from sqlalchemy.types import JSON
 
 from cidc_api.models.db.base_orm import BaseORM
 from cidc_api.models.types import (
-    PrimaryDiagnosisDiseaseGroup,
     TumorGrade,
     CancerStageSystem,
     CancerStageSystemVersion,
@@ -30,23 +29,23 @@ class DiseaseORM(BaseORM):
         "participant_id",
     ]
     __table_args__ = {"schema": "stage2"}
+    __data_category__ = "disease"
 
     disease_id: Mapped[int] = mapped_column(primary_key=True)
     participant_id: Mapped[int] = mapped_column(ForeignKey("stage2.participant.participant_id", ondelete="CASCADE"))
-    primary_diagnosis_disease_group: Mapped[PrimaryDiagnosisDiseaseGroup]
     primary_disease_site: Mapped[UberonAnatomicalTerm]
-    morphological_code: Mapped[ICDO3MorphologicalCode]
-    morphological_term: Mapped[ICDO3MorphologicalTerm]
+    morphological_code: Mapped[ICDO3MorphologicalCode | None]
+    morphological_term: Mapped[ICDO3MorphologicalTerm | None]
     cancer_type_description: Mapped[str | None]
-    days_since_original_diagnosis: Mapped[NonPositiveInt]
+    days_since_original_diagnosis: Mapped[NonPositiveInt | None]
     tumor_grade: Mapped[TumorGrade | None]
     cancer_stage_system: Mapped[CancerStageSystem]
     cancer_stage_system_version: Mapped[CancerStageSystemVersion | None] = mapped_column(String, nullable=True)
-    cancer_stage: Mapped[CancerStage] = mapped_column(String)
-    t_category: Mapped[TCategory]
-    n_category: Mapped[NCategory]
-    m_category: Mapped[MCategory]
-    metastatic_organ: Mapped[List[UberonAnatomicalTerm]] = mapped_column(JSON, nullable=True)
+    cancer_stage: Mapped[CancerStage | None] = mapped_column(String)
+    t_category: Mapped[TCategory | None]
+    n_category: Mapped[NCategory | None]
+    m_category: Mapped[MCategory | None]
+    metastatic_organ: Mapped[List[UberonAnatomicalTerm] | None] = mapped_column(JSON, nullable=True)
     solely_extramedullary_disease: Mapped[YNU]
     extramedullary_organ: Mapped[List[UberonAnatomicalTerm]] = mapped_column(JSON, nullable=True)
 

cidc_api/models/db/stage2/exposure_orm.py

@@ -11,11 +11,12 @@ class ExposureORM(BaseORM):
     __tablename__ = "exposure"
     __repr_attrs__ = ["exposure_id", "exposure_type", "carcinogen_exposure"]
     __table_args__ = {"schema": "stage2"}
+    __data_category__ = "exposure"
 
     exposure_id: Mapped[int] = mapped_column(primary_key=True)
     participant_id: Mapped[int] = mapped_column(ForeignKey("stage2.participant.participant_id", ondelete="CASCADE"))
 
     carcinogen_exposure: Mapped[YNU]
-    exposure_type: Mapped[ExposureType]
+    exposure_type: Mapped[ExposureType | None]
 
     participant: Mapped["ParticipantORM"] = relationship(back_populates="exposures", cascade="all, delete")

cidc_api/models/db/stage2/gvhd_diagnosis_acute_orm.py

@@ -19,6 +19,7 @@ class GVHDDiagnosisAcuteORM(BaseORM):
         {"schema": "stage2"},
     )
     __repr_attrs__ = ["gvhd_diagnosis_acute_id", "pre_or_post_enrollment"]
+    __data_category__ = "gvhd_diagnosis_acute"
 
     gvhd_diagnosis_acute_id: Mapped[int] = mapped_column(primary_key=True)
     participant_id: Mapped[int] = mapped_column(ForeignKey("stage2.participant.participant_id", ondelete="CASCADE"))

cidc_api/models/db/stage2/gvhd_diagnosis_chronic_orm.py

@@ -21,6 +21,7 @@ class GVHDDiagnosisChronicORM(BaseORM):
         {"schema": "stage2"},
     )
     __repr_attrs__ = ["gvhd_diagnosis_chronic_id", "pre_or_post_enrollment"]
+    __data_category__ = "gvhd_diagnosis_chronic"
 
     gvhd_diagnosis_chronic_id: Mapped[int] = mapped_column(primary_key=True)
     participant_id: Mapped[int] = mapped_column(ForeignKey("stage2.participant.participant_id", ondelete="CASCADE"))

cidc_api/models/db/stage2/gvhd_organ_acute_orm.py

@@ -9,6 +9,7 @@ class GVHDOrganAcuteORM(BaseORM):
     __tablename__ = "gvhd_organ_acute"
     __repr_attrs__ = ["gvhd_organ_acute_id", "organ"]
     __table_args__ = {"schema": "stage2"}
+    __data_category__ = "gvhd_organ_acute"
 
     gvhd_organ_acute_id: Mapped[int] = mapped_column(primary_key=True)
     gvhd_diagnosis_acute_id: Mapped[int] = mapped_column(

cidc_api/models/db/stage2/gvhd_organ_chronic_orm.py

@@ -9,6 +9,7 @@ class GVHDOrganChronicORM(BaseORM):
     __tablename__ = "gvhd_organ_chronic"
     __repr_attrs__ = ["gvhd_organ_chronic_id", "organ"]
     __table_args__ = {"schema": "stage2"}
+    __data_category__ = "gvhd_organ_chronic"
 
     gvhd_organ_chronic_id: Mapped[int] = mapped_column(primary_key=True)
     gvhd_diagnosis_chronic_id: Mapped[int] = mapped_column(

cidc_api/models/db/stage2/medical_history_orm.py

@@ -12,6 +12,7 @@ class MedicalHistoryORM(BaseORM):
     __tablename__ = "medical_history"
     __repr_attrs__ = ["medical_history_id"]
     __table_args__ = {"schema": "stage2"}
+    __data_category__ = "medical_history"
 
     medical_history_id: Mapped[int] = mapped_column(primary_key=True)
     participant_id: Mapped[int] = mapped_column(ForeignKey("stage2.participant.participant_id", ondelete="CASCADE"))

cidc_api/models/db/stage2/other_malignancy_orm.py

@@ -12,18 +12,19 @@ class OtherMalignancyORM(BaseORM):
     __tablename__ = "other_malignancy"
     __repr_attrs__ = ["other_malignancy_id", "primary_disease_site"]
     __table_args__ = {"schema": "stage2"}
+    __data_category__ = "other_malignancy"
 
     other_malignancy_id: Mapped[int] = mapped_column(primary_key=True)
     medical_history_id: Mapped[int] = mapped_column(
         ForeignKey("stage2.medical_history.medical_history_id", ondelete="CASCADE")
     )
 
-    primary_disease_site: Mapped[UberonAnatomicalTerm]
-    morphological_code: Mapped[Optional[ICDO3MorphologicalCode]]
-    morphological_term: Mapped[Optional[ICDO3MorphologicalTerm]]
-    malignancy_description: Mapped[Optional[str]]
-    days_since_diagnosis: Mapped[Optional[NonPositiveInt]]
-    malignancy_status: Mapped[Optional[MalignancyStatus]]
+    other_malignancy_primary_disease_site: Mapped[UberonAnatomicalTerm]
+    other_malignancy_morphological_code: Mapped[Optional[ICDO3MorphologicalCode]]
+    other_malignancy_morphological_term: Mapped[Optional[ICDO3MorphologicalTerm]]
+    other_malignancy_description: Mapped[Optional[str]]
+    other_malignancy_days_since_diagnosis: Mapped[Optional[NonPositiveInt]]
+    other_malignancy_status: Mapped[Optional[MalignancyStatus]]
 
     medical_history: Mapped["MedicalHistoryORM"] = relationship(
         back_populates="other_malignancies", cascade="all, delete"

cidc_api/models/db/stage2/participant_orm.py

@@ -4,7 +4,7 @@ from sqlalchemy import ForeignKey, ForeignKeyConstraint
 from sqlalchemy.orm import Mapped, mapped_column, relationship
 
 from cidc_api.models.db.base_orm import BaseORM
-from cidc_api.models.types import OffStudyReason
+from cidc_api.models.types import OffStudyReason, YNU
 
 
 class ParticipantORM(BaseORM):
@@ -16,17 +16,18 @@ class ParticipantORM(BaseORM):
         ),
         {"schema": "stage2"},
     )
+    __data_category__ = "participant"
 
     participant_id: Mapped[int] = mapped_column(primary_key=True)
     trial_id: Mapped[str]
     version: Mapped[str]
 
-    native_participant_id: Mapped[str]
-    cimac_participant_id: Mapped[str] = mapped_column(unique=True)
+    native_participant_id: Mapped[str | None]
+    cimac_participant_id: Mapped[str | None]
     consent_group_id: Mapped[Optional[int]] = mapped_column(
         ForeignKey("stage2.consent_group.consent_group_id", ondelete="CASCADE")
     )
-    off_study: Mapped[bool]
+    off_study: Mapped[YNU]
     off_study_reason: Mapped[Optional[OffStudyReason]]
     off_study_reason_other: Mapped[Optional[str]]
 

cidc_api/models/db/stage2/prior_treatment_orm.py

@@ -1,8 +1,9 @@
-from typing import Optional
+from typing import Optional, List
 
-from pydantic import NonPositiveInt, NonNegativeInt
+from pydantic import NonPositiveInt, NegativeInt
 from sqlalchemy import ForeignKey
 from sqlalchemy.orm import Mapped, mapped_column, relationship
+from sqlalchemy.types import JSON
 
 from cidc_api.models.db.base_orm import BaseORM
 from cidc_api.models.types import PriorTreatmentType, ConditioningRegimenType, StemCellDonorType
@@ -12,17 +13,18 @@ class PriorTreatmentORM(BaseORM):
     __tablename__ = "prior_treatment"
     __repr_attrs__ = ["prior_treatment_id", "type"]
     __table_args__ = {"schema": "stage2"}
+    __data_category__ = "prior_treatment"
 
     prior_treatment_id: Mapped[int] = mapped_column(primary_key=True)
     participant_id: Mapped[int] = mapped_column(ForeignKey("stage2.participant.participant_id", ondelete="CASCADE"))
 
-    days_to_start: Mapped[Optional[NonPositiveInt]]
-    days_to_end: Mapped[Optional[NonPositiveInt]]
-    type: Mapped[PriorTreatmentType]
-    description: Mapped[Optional[str]]
-    best_response: Mapped[Optional[str]]
-    conditioning_regimen_type: Mapped[Optional[ConditioningRegimenType]]
-    stem_cell_donor_type: Mapped[Optional[StemCellDonorType]]
-    days_from_transplant_to_treatment_initiation: Mapped[Optional[NonNegativeInt]]
+    prior_treatment_days_to_start: Mapped[Optional[NonPositiveInt]]
+    prior_treatment_days_to_end: Mapped[Optional[NonPositiveInt]]
+    prior_treatment_type: Mapped[List[PriorTreatmentType]] = mapped_column(JSON, nullable=False)
+    prior_treatment_description: Mapped[Optional[str]]
+    prior_treatment_best_response: Mapped[Optional[str]]
+    prior_treatment_conditioning_regimen_type: Mapped[Optional[ConditioningRegimenType]]
+    prior_treatment_stem_cell_donor_type: Mapped[Optional[StemCellDonorType]]
+    prior_treatment_days_to_prior_transplant: Mapped[Optional[NegativeInt]]
 
     participant: Mapped["ParticipantORM"] = relationship(back_populates="prior_treatments", cascade="all, delete")

cidc_api/models/db/stage2/radiotherapy_dose_orm.py

@@ -18,6 +18,7 @@ class RadiotherapyDoseORM(BaseORM):
     __tablename__ = "radiotherapy_dose"
     __repr_attrs__ = ["radiotherapy_dose_id", "procedure"]
     __table_args__ = {"schema": "stage2"}
+    __data_category__ = "radiotherapy_dose"
 
     radiotherapy_dose_id: Mapped[int] = mapped_column(primary_key=True)
     treatment_id: Mapped[int] = mapped_column(ForeignKey("stage2.treatment.treatment_id", ondelete="CASCADE"))

cidc_api/models/db/stage2/response_by_system_orm.py

@@ -10,6 +10,7 @@ class ResponseBySystemORM(BaseORM):
     __tablename__ = "response_by_system"
     __repr_attrs__ = ["response_by_system_id", "participant_id"]
     __table_args__ = {"schema": "stage2"}
+    __data_category__ = "response_by_system"
 
     response_by_system_id: Mapped[int] = mapped_column(primary_key=True)
     participant_id: Mapped[int] = mapped_column(ForeignKey("stage2.participant.participant_id", ondelete="CASCADE"))
@@ -17,6 +18,7 @@ class ResponseBySystemORM(BaseORM):
     response_system_version: Mapped[ResponseSystemVersion] = mapped_column(String)
     best_overall_response: Mapped[BestOverallResponse] = mapped_column(String)
     response_duration: Mapped[PositiveInt | None]
+    duration_of_stable_disease: Mapped[PositiveInt | None]
     durable_clinical_benefit: Mapped[bool | None]
     days_to_first_response: Mapped[PositiveInt | None]
     days_to_best_response: Mapped[PositiveInt | None]

cidc_api/models/db/stage2/response_orm.py

@@ -11,17 +11,18 @@ class ResponseORM(BaseORM):
     __tablename__ = "response"
     __repr_attrs__ = ["response_id", "participant_id"]
     __table_args__ = {"schema": "stage2"}
+    __data_category__ = "response"
 
     response_id: Mapped[int] = mapped_column(primary_key=True)
     participant_id: Mapped[int] = mapped_column(ForeignKey("stage2.participant.participant_id", ondelete="CASCADE"))
     survival_status: Mapped[SurvivalStatus]
-    overall_survival: Mapped[NonNegativeInt | None]
+    overall_survival: Mapped[NonNegativeInt]
     abscopal_response: Mapped[YNUNA | None]
     pathological_complete_response: Mapped[YNUNA | None]
     days_to_death: Mapped[NonNegativeInt | None]
     cause_of_death: Mapped[CauseOfDeath | None]
     evaluable_for_toxicity: Mapped[bool]
     evaluable_for_efficacy: Mapped[bool]
-    days_to_last_follow_up: Mapped[NonNegativeInt | None]
+    days_to_last_vital_status: Mapped[NonNegativeInt | None]
 
     participant: Mapped["ParticipantORM"] = relationship(back_populates="response", cascade="all, delete")

cidc_api/models/db/stage2/stem_cell_transplant_orm.py

@@ -12,6 +12,7 @@ class StemCellTransplantORM(BaseORM):
     __tablename__ = "stem_cell_transplant"
     __repr_attrs__ = ["stem_cell_transplant_id"]
     __table_args__ = {"schema": "stage2"}
+    __data_category__ = "stem_cell_transplant"
 
     stem_cell_transplant_id: Mapped[int] = mapped_column(primary_key=True)
     treatment_id: Mapped[int] = mapped_column(ForeignKey("stage2.treatment.treatment_id", ondelete="CASCADE"))

cidc_api/models/db/stage2/surgery_orm.py

@@ -12,6 +12,7 @@ class SurgeryORM(BaseORM):
     __tablename__ = "surgery"
     __repr_attrs__ = ["surgery_id", "procedure"]
     __table_args__ = {"schema": "stage2"}
+    __data_category__ = "surgery"
 
     surgery_id: Mapped[int] = mapped_column(primary_key=True)
     treatment_id: Mapped[int] = mapped_column(ForeignKey("stage2.treatment.treatment_id", ondelete="CASCADE"))

cidc_api/models/db/stage2/therapy_agent_dose_orm.py

@@ -12,6 +12,7 @@ class TherapyAgentDoseORM(BaseORM):
     __tablename__ = "therapy_agent_dose"
     __repr_attrs__ = ["therapy_agent_dose_id", "therapy_agent_name"]
     __table_args__ = {"schema": "stage2"}
+    __data_category__ = "therapy_agent_dose"
 
     therapy_agent_dose_id: Mapped[int] = mapped_column(primary_key=True)
     treatment_id: Mapped[int] = mapped_column(ForeignKey("stage2.treatment.treatment_id", ondelete="CASCADE"))

cidc_api/models/db/stage2/treatment_orm.py

@@ -11,6 +11,7 @@ class TreatmentORM(BaseORM):
     __tablename__ = "treatment"
     __repr_attrs__ = ["treatment_id", "participant_id", "treatment_description"]
     __table_args__ = {"schema": "stage2"}
+    __data_category__ = "treatment"
 
     treatment_id: Mapped[int] = mapped_column(primary_key=True)
     participant_id: Mapped[int] = mapped_column(ForeignKey("stage2.participant.participant_id", ondelete="CASCADE"))

cidc_api/models/db/stage2/trial_orm.py

@@ -6,35 +6,38 @@ from sqlalchemy.orm import Mapped, mapped_column, relationship
 from sqlalchemy.types import JSON
 
 from cidc_api.models.db.base_orm import BaseORM
-from cidc_api.models.types import TrialStatus, AssayType, TrialOrganization, TrialFundingAgency
+from cidc_api.models.types import AssayType, TrialOrganization, TrialFundingAgency, AgeGroup
 
 
 class TrialORM(BaseORM):
     __tablename__ = "trial"
     __repr_attrs__ = ["trial_id", "version"]
     __table_args__ = {"schema": "stage2"}
+    __data_category__ = "study"
 
     trial_id: Mapped[str] = mapped_column(primary_key=True)
     version: Mapped[str] = mapped_column(primary_key=True)
 
-    nct_id: Mapped[Optional[str]]
-    nci_id: Mapped[Optional[str]]
-    trial_name: Mapped[Optional[str]]
-    trial_type: Mapped[Optional[str]]
-    trial_description: Mapped[Optional[str]]
-    trial_organization: Mapped[Optional[TrialOrganization]]
-    grant_or_affiliated_network: Mapped[Optional[TrialFundingAgency]]
-    trial_status: Mapped[TrialStatus]
-    biobank_institution_id: Mapped[Optional[int]]
-    justification: Mapped[Optional[str]]
+    primary_endpoint: Mapped[str | None]
+    age_group: Mapped[List[AgeGroup]] = mapped_column(JSON, nullable=True)
+    study_population: Mapped[str | None]
+    nct_id: Mapped[str | None]
+    nci_id: Mapped[str | None]
+    trial_name: Mapped[str | None]
+    trial_type: Mapped[str | None]
+    trial_description: Mapped[str | None]
+    trial_organization: Mapped[TrialOrganization | None]
+    grant_or_affiliated_network: Mapped[TrialFundingAgency | None]
+    biobank_institution_id: Mapped[int | None]
+    justification: Mapped[str | None]
     dates_of_conduct_start: Mapped[datetime]
-    dates_of_conduct_end: Mapped[Optional[datetime]]
-    schema_file_id: Mapped[Optional[int]]
-    biomarker_plan: Mapped[Optional[str]]
-    data_sharing_plan: Mapped[Optional[str]]
-    expected_assays: Mapped[Optional[List[AssayType]]] = mapped_column(JSON, nullable=True)
+    dates_of_conduct_end: Mapped[datetime | None]
+    schema_file_id: Mapped[int | None]
+    biomarker_plan: Mapped[str | None]
+    data_sharing_plan: Mapped[str | None]
+    expected_assays: Mapped[List[AssayType]] = mapped_column(JSON, nullable=True)
     is_liquid_tumor_trial: Mapped[bool]
-    dbgap_study_accession: Mapped[Optional[str]]
+    dbgap_study_accession: Mapped[str | None]
 
     biobank: Mapped["InstitutionORM"] = relationship(back_populates="trial")
     schema: Mapped[Optional["FileORM"]] = relationship(back_populates="trial", viewonly=True)

cidc_api/models/models.py

@@ -3399,6 +3399,7 @@ class PreprocessedFiles(CommonColumns):
                 (cls.file_category == latest_subquery.c.file_category)
                 & (cls._created == latest_subquery.c.latest_created),
             )
+            .order_by(cls.file_category)
             .all()
         )
         return latest_files

cidc_api/models/pydantic/stage2/additional_treatment.py

@@ -10,14 +10,16 @@ class AdditionalTreatment(Base):
     additional_treatment_id: int | None = None
 
     # The unique internal identifier for the associated Participant record
-    participant_id: int | None = None
+    participant_id: str | None = (
+        None  # TODO: fix all stage2 models for participant_id to not be nullable, after s1 models are complete
+    )
 
     # Number of days from the enrollment date to the first recorded administration or occurrence of the treatment modality.
-    days_to_start: NonNegativeInt | None = None
+    additional_treatment_days_to_start: NonNegativeInt | None = None
 
     # Number of days from the enrollment date to the last recorded administration or occurrence of the treatment modality.
-    days_to_end: NonNegativeInt | None = None
+    additional_treatment_days_to_end: NonNegativeInt | None = None
 
     # Description of the prior treatment such as its full generic name if it is a type of therapy agent, radiotherapy procedure
     # name and location, or surgical procedure name and location.
-    description: str
+    additional_treatment_description: str

cidc_api/models/pydantic/stage2/adverse_event.py

@@ -25,7 +25,7 @@ class AdverseEvent(Base):
     adverse_event_id: int | None = None
 
     # The unique internal identifier of the associated participant
-    participant_id: int | None = None
+    participant_id: str | None = None
 
     # The unique internal identifier of the attributed treatment, if any
     treatment_id: int | None = None

cidc_api/models/pydantic/stage2/base.py

@@ -6,7 +6,11 @@ import copy
 
 class Base(BaseModel):
 
-    model_config = ConfigDict(validate_assignment=True, from_attributes=True)
+    model_config = ConfigDict(
+        validate_assignment=True,
+        from_attributes=True,
+        # extra="forbid" # TODO: enable me one stage1 models are implemented
+    )
 
     # Validates the new state and updates the object if valid
     def update(self, **kwargs):
@@ -28,3 +32,17 @@ class Base(BaseModel):
         except:
             self.__dict__.update(original_dict)
             raise
+
+    @classmethod
+    def split_list(cls, val):
+        """Listify fields that are multi-valued in input data, e.g. 'lung|kidney'"""
+        if type(val) == list:
+            return val
+        elif type(val) == str:
+            if not val:
+                return []
+            return val.split("|")
+        elif val == None:
+            return []
+        else:
+            raise ValueError("Field value must be string or list")

cidc_api/models/pydantic/stage2/baseline_clinical_assessment.py

@@ -10,7 +10,7 @@ class BaselineClinicalAssessment(Base):
     baseline_clinical_assessment_id: int | None = None
 
     # The unique identifier for the associated participant
-    participant_id: int | None = None
+    participant_id: str | None = None
 
     # The numerical score that represents the functional capabilities of a participant at the
     # enrollment date using the Eastern Cooperative Oncology Group Performance Status assessment.

cidc_api/models/pydantic/stage2/demographic.py

@@ -1,6 +1,6 @@
-from typing import Self
+from typing import Self, Annotated, List
 
-from pydantic import PositiveInt, NonNegativeFloat, PositiveFloat, model_validator, field_validator
+from pydantic import PositiveInt, NonNegativeFloat, PositiveFloat, model_validator, field_validator, BeforeValidator
 
 from .base import Base
 from cidc_api.models.types import (
@@ -25,7 +25,7 @@ class Demographic(Base):
 
     # The unique internal identifier for the associated participant
     # CDE: https://cadsr.cancer.gov/onedata/dmdirect/NIH/NCI/CO/CDEDD?filter=CDEDD.ITEM_ID=12220014%20and%20ver_nr=1
-    participant_id: int | None = None
+    participant_id: str | None = None
 
     # The age of the subject when the subject enrolled in the study.
     # CDE: https://cadsr.cancer.gov/onedata/dmdirect/NIH/NCI/CO/CDEDD?filter=CDEDD.ITEM_ID=15742605%20and%20ver_nr=1
@@ -34,9 +34,9 @@ class Demographic(Base):
     # Unit of measurement for the age of the participant. e.g. "Years"
     age_at_enrollment_units: AgeAtEnrollmentUnits | None = None
 
-    # Indicates whether the participant is 90 years old or older. (for PHI purposes)
+    # Indicates whether the participant is 90 years old or over. (for PHI purposes)
     # CDE: https://cadsr.cancer.gov/onedata/dmdirect/NIH/NCI/CO/CDEDD?filter=CDEDD.ITEM_ID=15354920%20and%20ver_nr=1
-    age_90_or_older: bool
+    age_90_or_over: bool
 
     # A textual description of a person's sex at birth. e.g. "Male"
     # CDE: https://cadsr.cancer.gov/onedata/dmdirect/NIH/NCI/CO/CDEDD?filter=CDEDD.ITEM_ID=7572817%20and%20ver_nr=3
@@ -44,7 +44,7 @@ class Demographic(Base):
 
     # The race of the participant. e.g. "White"
     # CDE: https://cadsr.cancer.gov/onedata/dmdirect/NIH/NCI/CO/CDEDD?filter=CDEDD.ITEM_ID=2192199%20and%20ver_nr=1
-    race: Race
+    race: Annotated[List[Race], BeforeValidator(Base.split_list)]
 
     # The ethnicity of the participant. e.g. "Hispanic or Latino"
     # CDE: https://cadsr.cancer.gov/onedata/dmdirect/NIH/NCI/CO/CDEDD?filter=CDEDD.ITEM_ID=2192217%20and%20ver_nr=2
@@ -68,15 +68,15 @@ class Demographic(Base):
 
     # The body mass index of the participant at the enrollment date.
     # CDE: https://cadsr.cancer.gov/onedata/dmdirect/NIH/NCI/CO/CDEDD?filter=CDEDD.ITEM_ID=2006410%20and%20ver_nr=3
-    body_mass_index: PositiveFloat
+    body_mass_index: PositiveFloat | None = None
 
     # A decimal number that represents the measure of the 2-dimensional extent of the body surface (i.e., the skin) of the participant at the enrollment date.
     # CDE: https://cadsr.cancer.gov/onedata/dmdirect/NIH/NCI/CO/CDEDD?filter=CDEDD.ITEM_ID=6606197%20and%20ver_nr=1
-    body_surface_area: PositiveFloat
+    body_surface_area: PositiveFloat | None = None
 
     # Unit of measurement for body surface area of the participant at the enrollment date. e.g. "m2"
     # https://cadsr.cancer.gov/onedata/dmdirect/NIH/NCI/CO/CDEDD?filter=CDEDD.ITEM_ID=15114329%20and%20ver_nr=1
-    body_surface_area_units: BodySurfaceAreaUnits
+    body_surface_area_units: BodySurfaceAreaUnits | None = None
 
     # The occupation/job category of the participant. e.g. "Manager"
     # CDE: https://cadsr.cancer.gov/onedata/dmdirect/NIH/NCI/CO/CDEDD?filter=CDEDD.ITEM_ID=6617540%20and%20ver_nr=1
@@ -92,23 +92,32 @@ class Demographic(Base):
 
     @model_validator(mode="after")
     def validate_age_at_enrollment_cr(self) -> Self:
-        if self.age_90_or_older:
+        if self.age_90_or_over:
             if self.age_at_enrollment or self.age_at_enrollment_units:
                 raise ValueError(
-                    'If "age_90_or_older" is "Yes" then "age_at_enrollment" and "age_at_enrollment_units" must be blank.'
+                    'If "age_90_or_over" is "Yes" then "age_at_enrollment" and "age_at_enrollment_units" must be blank.'
                 )
         else:
             if not self.age_at_enrollment or not self.age_at_enrollment_units:
                 raise ValueError(
-                    'If "age_90_or_older" is "No" then "age_at_enrollment" and "age_at_enrollment_units" are required.'
+                    'If "age_90_or_over" is "No" then "age_at_enrollment" and "age_at_enrollment_units" are required.'
                 )
         return self
 
     @model_validator(mode="after")
     def validate_age_at_enrollment_value(self) -> Self:
-        age_in_years = (
-            self.age_at_enrollment if self.age_at_enrollment_units == "Years" else self.age_at_enrollment / 365.25
-        )
-        if age_in_years >= 90:
-            raise ValueError('"age_at_enrollment" cannot represent a value greater than 90 years of age.')
+        if not self.age_90_or_over:
+            age_in_years = (
+                self.age_at_enrollment if self.age_at_enrollment_units == "Years" else self.age_at_enrollment / 365.25
+            )
+            if age_in_years >= 90:
+                raise ValueError('"age_at_enrollment" cannot represent a value greater than 90 years of age.')
+        return self
+
+    @model_validator(mode="after")
+    def validate_body_surface_area_units_cr(self) -> Self:
+        if self.body_surface_area and not self.body_surface_area_units:
+            raise ValueError(
+                'If "body_surface_area" is provided then "body_surface_area_units_other" must also be provided.'
+            )
         return self