include_access_model.yaml

---
id: https://includedcc.org/include-access-model
name: include-access-model
title: INCLUDE DCC Access Model
description: LinkML Schema for the internal INCLUDE DCC Access Model
license: MIT
see_also:
  - https://includedcc.github.io/include-access-model

prefixes:
  includedcc: https://includedcc.org/include-access-model/
  linkml: https://w3id.org/linkml/
  schema: http://schema.org/
  # subset of CDC Race and Ethnicity Code Set Version 1.0
  cdc_race_eth: urn:oid:2.16.840.1.113883.6.238/
  hl7_null: http://terminology.hl7.org/CodeSystem/v3-NullFlavor/
  ig_dob_method: https://nih-ncpi.github.io/ncpi-fhir-ig-2/CodeSystem/research-data-date-of-birth-method/
  igcondtype: https://nih-ncpi.github.io/ncpi-fhir-ig-2/CodeSystem/condition-type/
  ig2dac: https://nih-ncpi.github.io/ncpi-fhir-ig-2/CodeSystem/research-data-access-code/
  ig2dat: https://nih-ncpi.github.io/ncpi-fhir-ig-2/CodeSystem/research-data-access-type/
  ig2_biospecimen_availability: https://nih-ncpi.github.io/ncpi-fhir-ig-2/CodeSystem/biospecimen-availability/
  snomed_ct: 	http://snomed.info/id/
  MONDO: http://purl.obolibrary.org/obo/MONDO_
  HP: http://purl.obolibrary.org/obo/HP_
  mesh: http://id.nlm.nih.gov/mesh/
  NCIT: http://purl.obolibrary.org/obo/NCIT_
  PATO: http://purl.obolibrary.org/obo/PATO_

default_prefix: includedcc
default_range: string

imports:
  - linkml:types

classes:
  Record:
    description: One row / entity within the database
    title: Record
    abstract: true
    slots:
    - external_id
  Study:
    title: Research Study 
    description: Study Metadata 
    is_a: Record
    slots: 
    #TODO: Split out core Study items and additional study metadata?
      - study_id
      - parent_study 
      - study_title
      - study_code
      - study_short_name
      - program
      - funding_source
      - principal_investigator 
      - contact
      - study_description
      - website
    # - dbgap : Should we call this out specifically or just use an external id?
      - publication
      - acknowledgments
      - citation_statement
      - do_id
    slot_usage:
      study_id:
        range: string
        required: true
        identifier: true
  StudyMetadata:
    title: Study Metadata
    description: Additional features about studies that may not apply to all studies
    is_a: Record
    slots:
      - study_id
      - participant_lifespan_stage
      - selection_criteria
      - study_design
      - clinical_data_source_type
      - data_category
      - vbr
      - research_domain
      - expected_number_of_participants
      - actual_number_of_participants
      # Do we need this info, or should it be documented elsewhere?
      #- guidType
      #- guidMapped
    slot_usage:
      study_id:
        required: true
        identifier: true
      data_category:
        required: true
        multivalued: true

  VirtualBiorepository:
    title: Virtual BioRepository (VBR)
    description: An organization that can provide access to specimen for further analysis.
    is_a: Record
    slots:
      - name
      - institution
      - contact
      - website
      - vbr_readme
  DOI:
    title: Digital Object Identifier (DOI)
    description: A DOI is a permanent reference with metadata about a digital object.
    is_a: Record
    slots:
      - do_id
      - bibliographic_reference
    slot_usage:
      do_id:
        range: string
        required: true
        identifier: true
  Investigator:
    title: Investigator
    description: An individual who made contributions to the collection, analysis, or sharing of data.
    is_a: Record
    slots:
      - name
      - institution
      # - orcid : Should we call this out specifically or just use an external id?
      - investigator_title
      - email
  Publication:
    title: Publication
    description: Information about a specific publication.
    is_a: Record
    #Could add more here
    slots:
      - bibliographic_reference
      - website
  Subject:
    title: Subject
    description: This entity is the subject about which data or references are recorded. 
      This includes the idea of a human participant in a study, a cell line, an animal model, 
      or any other similar entity.
    is_a: Record
    slots:
      - subject_id
      - subject_type
      - organism_type
    slot_usage:
      subject_id:
        range: string
        required: true
        identifier: true
  Demographics:
    title: Demographics
    description: Basic participant demographics summary
    is_a: Record
    slots:
      - subject_id
      - sex
      - race
      - ethnicity
      - down_syndrome_status
      - age_at_last_vital_status 
      - vital_status
      - age_at_first_engagement
    slot_usage:
      subject_id:
        required: true
        identifier: true
  SubjectAssertion:
    title: Subject Assertion 
    description: Assertion about a particular Subject. May include Conditions, Measurements, etc.
    is_a: Record 
    slots:
      - assertion_id
      - subject_id
      - encounter_id
      - assertion_provenance
      - age_at_assertion
      - age_at_event
      - age_at_resolution
      - concept
      - concept_source
      - value_concept
      - value_number
      - value_source
      - value_unit
      - value_unit_source
    slot_usage:
      assertion_id:
        range: string
        required: true
        identifier: true
  Concept:
    title: Concept
    description: A standardized concept with display information.
    slots:
      - concept_curie
      - display
    slot_usage:
      concept_curie:
        required: true
        identifier: true
  Sample:
    title: Sample 
    description: A functionally equivalent specimen taken from a participant or processed from such a sample.
    is_a: Record
    slots: 
      - sample_id
      - biospecimen_collection_id 
      - parent_sample_id 
      - sample_type
      - processing 
      - availablity_status 
      - storage_method 
      - quantity_number
      - quantity_unit 
    slot_usage:
      sample_id:
        range: string
        required: true
        identifier: true
      biospecimen_collection_id:
        description: Biospecimen Collection during which this sample was generated.
  BiospecimenCollection:
    title: BiospecimenCollection
    description: A biospecimen collection event which yields one or more Samples.
    is_a: Record 
    slots: 
      - biospecimen_collection_id
      - age_at_collection
      - method 
      - site 
      - spatial_qualifier 
      - laterality 
      - encounter_id
    slot_usage:
      biospecimen_collection_id:
        range: string
        required: true
        identifier: true
  Aliquot:
    title: Aliquot
    description: A specific tube or amount of a biospecimen associated with a Sample.
    is_a: Record 
    slots: 
      - aliquot_id
      - sample_id
      - availablity_status
      - quantity_number
      - quantity_unit 
      - concentration_number
      - concentration_unit
    slot_usage:
      aliquot_id:
        range: string
        required: true
        identifier: true
  Encounter:
    title: Participant Encounter
    description: An event at which data was collected about a participant, 
      an intervention was made, or information about a participant was recorded.
    is_a: Record
    slots:
      - encounter_id
      - subject_id
      #TODO: Should this alternatively be an "encounter type", ie, to not require a full definition?
      - encounter_definition_id
      - age_at_event
    slot_usage:
      encounter_id:
        range: string
        required: true
        identifier: true
  EncounterDefinition:
    title: Encounter Definition
    description: A definition of an encounter type in this study, ie, 
      an event at which data was collected about a participant, 
      an intervention was made, or information about a participant was recorded.
      This may be something planned by a study or a type of data collection.
    #TODO: These are metadata and may not need the same Record basis.
    is_a: Record
    slots:
      - encounter_definition_id
      - name
      - description
      - activity_definition_id
    slot_usage:
      encounter_definition_id:
        range: string
        required: true
        identifier: true
      activity_definition_id:
        multivalued: true
  ActivityDefinition:
    title: Activity Definition
    description: A definition of an activity in this study, eg, 
      a biospecimen collection, intervention, survey, or assessment.
    #TODO: These are metadata and may not need the same Record basis.
    is_a: Record
    slots:
      - activity_definition_id
      - name
      - description
      #TODO: Probably want an "expected data generated" slot, eg, 
      #observation definitions or dd refs
    slot_usage:
      activity_definition_id:
        range: string
        required: true
        identifier: true
  File:
    title: File
    description: File
    is_a: Record 
    slots: 
      - file_id
      - subject_id
      - sample_id
      - filename
      - format
      - data_category
      - data_type
      - format
      - size
      #TODO: I'm not convinced this is the right strategy- access model vs operations
      - staging_url
      - release_url
      - drs_uri
      - hash
    slot_usage:
      file_id:
        range: string
        required: true
        identifier: true
      subject_id:
        multivalued: true
      sample_id:
        multivalued: true
  FileHash:
    title: File Hash
    description: Type and value of a file content hash.
    slots:
      - hash_type
      - hash_value
  Dataset:
    title: Dataset
    description: Set of files grouped together for release.
    slots:
      - dataset_id
      - name
      - description
      - do_id
      - file_id
      - publication
      #TODO: Are these good elements for the core entity?
      - data_collection_start
      - data_collection_end
      
    slot_usage:
      dataset_id:
        range: string
        required: true
        identifier: true
      file_id: 
        multivalued: true
        description: The list of files comprising this dataset.


slots: 
  study_id:
    title: Study ID
    description: INCLUDE Global ID for the study
    range: Study
    multivalued: false 
  do_id:
    title: DOI
    description: Digital Object Identifier (DOI) for this Record.
    range: DOI
    multivalued: false 
  subject_id:
    title: Study ID
    description: INCLUDE Global ID for the Subject
    range: Subject
    multivalued: false 
  assertion_id:
    title: Assertion ID
    description: INCLUDE Global ID for the Assertion
    range: SubjectAssertion
    multivalued: false 
  external_id: 
    title: External Identifiers
    description: Other identifiers for this entity, eg, from the submitting study or in systems like dbGaP
    required: false 
    range: uriorcurie 
    multivalued: true 
  parent_study:
    title: Parent Study
    description: The parent study for this study, if it is a nested study.
    required: false 
    range: Study 
    multivalued: false
  funding_source:
    title: Funding Source 
    description: The funding source(s) of the study.
    required: false 
    range: string 
    multivalued: true 
  principal_investigator:  
    title: Principal Investigator 
    description: The Principal Investigator(s) responsible for the study.
    required: true 
    range: Investigator
    multivalued: true 
  study_title: 
    description: Full Study Title  
    required: true 
    range: string 
    multivalued: false 
  study_code:
    description: Unique identifier for the study (generally a short acronym)
    title: Study Code
    range: string
    required: true
  study_short_name:
    title: Study Code
    description: Short name for the study
    range: string
    required: false
  investigator_title:
    title: Investigator Title
    description: The title of the Investigator, eg, "Assistant Professor"
    range: string
    required: false
  name:
    title: Name
    description: Name of the entity.
    range: string
    required: false
  email:
    title: Email Address
    description: An email address to reach the entity.
    range: string
    required: false
  institution:
    title: Institution
    description: Name of the institution this record is associated with.
    range: string
    required: false
  program:
    title: Program
    description: Funding source(s) for the study
    range: EnumProgram
    required: true
    multivalued: true
  study_description:
    title: Study Description
    description: Brief description of the study (2-4 sentences)
    range: string
    required: true
  website:
    title: Website
    description: Website for the Record.
    range: uri
  contact:
    title: Contact Person
    description: The individual to contact with questions about this record.
    required: true 
    range: Investigator
    multivalued: true 
  vbr:
    title: Virtual Biorepository
    description: Information about the study's Virtual Biorepository, if participating
    range: VirtualBiorepository
    required: false
  vbr_readme:
    title: VBR Readme
    description: Instructions for contacting or requesting samples from Virtual Biorepository, if participating
    range: string
  research_domain:
    description: Main research domain(s) of the study, other than Down syndrome
    range: EnumResearchDomain
    required: true
    multivalued: true
  participant_lifespan_stage:
    title: Participant Lifespan Stage
    description: Focus age group(s) of the study population
    range: EnumParticipantLifespanStage
    required: true
    multivalued: true
  selection_criteria:
    title: Selection Criteria
    description: Brief description of inclusion and/or exclusion criteria for the study
    range: string
  study_design:
    title: Study Design
    description: Overall design of study, including whether it is longitudinal and whether family members/unrelated controls are also enrolled
    range: EnumStudyDesign
    required: true
    multivalued: true
  data_category:
    title: Data Category
    description: General category of data in this Record (e.g. Clinical, Genomics, etc)
    range: EnumDataCategory
  clinical_data_source_type:
    title: Clinical Data Source Type
    description: Source(s) of data collected from study participants
    range: EnumClinicalDataSourceType
    required: true
    multivalued: true
  publication:
    title: Publication
    description: Publications associated with this Record.
    range: Publication
    multivalued: true
  expected_number_of_participants:
    title: Expected Number of Participants
    description: Total expected number of participants to be recruited.
    range: integer
    required: true
  actual_number_of_participants:
    title: Actual Number of Participants
    description: Total participants included at this time.
    range: integer
    required: true
  acknowledgments:
    title: Acknowledgments
    description: Funding statement and acknowledgments for this study
    range: string
  citation_statement:
    title: Citation Statement
    description: Statement that secondary data users should use to acknowledge use of this study or dataset. E.g., "The results analyzed and <published or shown> here are based in whole or in part upon data generated by the INCLUDE (INvestigation of Co-occurring conditions across the Lifespan to Understand Down syndromE) Project <insert accession number(s) and/or study DOI(s)>, and were accessed from the INCLUDE Data Hub and <insert other database(s)>."
    range: string
  bibliographic_reference:
    title: Bibiliographic Reference
    description: Text use to reference this Record.
    range: string
  organism_type:
    title: Organism Type 
    description: Organism Type
    required: false 
    range: uriorcurie 
  subject_type: 
    title: Subject Type 
    description: Type of entity this record represents
    required: true 
    range: EnumSubjectType 
  sex:
    title: Sex
    description: Sex of Participant
    range: EnumSex
    required: true
  race:
    title: Race
    description: Race of Participant
    range: EnumRace
    required: true
    multivalued: true
  ethnicity:
    title: Ethnicity
    description: Ethnicity of Participant
    range: EnumEthnicity
    required: true
  down_syndrome_status:
    title: Down Syndrome Status
    description: Down Syndrome status of participant
    range: EnumDownSyndromeStatus
    required: true
  age_at_first_engagement:
    #Should this just be a reference out to an encounter?
    title: Age at First Participant Engagement
    description: Age in days of Participant at first recorded study event (enrollment, visit, observation, sample collection, survey completion, etc.). Age at enrollment is preferred, if available.
    range: integer
    unit:
      ucum_code: d
    #Range -1 year to 89*365.25 (Privacy upper bound for ages over 89 years)
    minimum_value: -365
    maximum_value: 32507    
  vital_status:
    title: Vital Status
    description: Whether participant is alive or dead
    range: EnumVitalStatus
  age_at_last_vital_status:
    title: Age at Last Vital Status
    description: Age in days when participant's vital status was last recorded
    range: integer
    unit:
      ucum_code: d
    minimum_value: -365
    maximum_value: 32507
  assertion_provenance:
    title: Assertion Provenance
    description: The original source of this assertion
    range: EnumAssertionProvenance
  age_at_assertion:
    title: Age at assertion
    description: The age in days of the Subject when the assertion was made.
    range: integer
    unit:
      ucum_code: d
  age_at_event:
    title: Age at event
    description: The age in days of the Subject at the time point which the assertion describes,
      eg, age of onset or when a measurement was performed.
    range: integer
    unit:
      ucum_code: d
  age_at_resolution:
    title: Age at resolution
    description: The age in days of the Subject when the asserted state was resolved.
    range: integer
    unit:
      ucum_code: d
  concept:
    title: Concept
    description: The structured term defining the meaning of the assertion.
    range: Concept
    multivalued: true
  concept_curie:
    title: Concept Curie
    description: The standardized curie for the term.
    range: uriorcurie
  display:
    title: Display String
    description: The friendly display string of the coded term.
    range: string
  concept_source:
    title: Concept Source Text
    description: The source text yielding the standardized concept.
    range: string
  value_concept:
    title: Value concept
    description: The structured term defining the value of the assertion.
    range: Concept
    multivalued: true
  value_number:
    title: Value Number
    description: The numeric value of the assertion.
    range: float
  value_source:
    title: Value Source Text
    description: The source text yielding the value.
    range: string
  value_unit:
    title: Value Units
    description: The structured term defining the units of the value.
    range: Concept
  value_unit_source:
    title: Value Units Source Text
    description: The source text yielding the value's units.
    range: string
  sample_id:
    title: Sample ID
    description: The unique identifier for this Sample.
    range: Sample
  parent_sample_id: 
    title: Parent Sample ID
    description: Sample from which this sample is derived
    range: Sample 
    inlined: false
  biospecimen_collection_id: 
    title: Biospecimen Collection ID
    description: Unique identifier for this Biospecimen Collection.
    range: BiospecimenCollection 
  aliquot_id: 
    title: Aliquot ID
    description: Unique identifier for an Aliquot.
    range: Aliquot 
  sample_type:
    title: Sample Type
    description: Type of material of which this Sample is comprised 
    required: true 
    range: uriorcurie  
  processing:
    title: Sample Processing
    description: Processing that was applied to the Parent Sample or from the Biospecimen Collection that yielded this distinct sample
    range: uriorcurie 
    multivalued: true 
  availablity_status: 
    title: Sample Availability
    description: Can this Sample be requested for further analysis?
    range: EnumAvailabilityStatus 
  storage_method: 
    title: Sample Storage Method
    description: Sample storage method, eg, Frozen or with additives
    range: uriorcurie
    multivalued: true
  quantity_number:
    title: Quantity
    description: The total quantity of the specimen
    range: float
  quantity_unit:
    title: Quantity Units
    description: The structured term defining the units of the quantity.
    range: Concept
  concentration_number: 
    title: Concentration
    description: What is the concentration of the analyte in the Aliquot?
    range: float 
  concentration_unit: 
    title: Concentration Units
    description: Units associated with the concentration of the analyte in the Aliquot.
    range: Concept 
  age_at_collection:
    title: Age at Biospecimen Collection
    description: The age at which this biospecimen was collected in decimal years.
    range: float 
    unit:
      ucum_code: a   
  method: 
    title: Biospecimen Collection Method 
    description: The approach used to collect the biospecimen.
    range: EnumSampleCollectionMethod
  site: 
    title: Biospecimen Collection Site  
    description: The location of the specimen collection.
    range: EnumSite
  spatial_qualifier: 
    title: Spatial Qualifier
    description: Qualifier that further refine the specific location of biospecimen collection
    range: EnumSpatialQualifiers
  laterality: 
    title: Location Laterality
    description: Laterality that further refine the specific location of biospecimen collection
    range: EnumLaterality
  encounter_id:
    title: Encounter ID
    description: Unique identifier for this Encounter.
    range: Encounter
  description:
    title: Description
    description: Description for this entity.
    range: string
  encounter_definition_id:
    title: Encounter Definition ID
    description: Unique identifier for this Encounter Definition.
    range: EncounterDefinition
  activity_definition_id:
    title: Activity Definition ID
    description: Unique identifier for this Activity Definition.
    range: ActivityDefinition
  file_id:
    title: File ID
    description: Unique identifier for this File.
    range: File
  filename:
    title: Filename
    description: The name of the file.
    range: string
  format:
    title: File Format
    description: The format of the file.
    range: EnumEDAMFormats
  data_type:
    title: Data Type
    description: The type of data within this file.
    range: EnumEDAMDataTypes
  size:
    title: File Size
    description: Size of the file, in Bytes.
    range: integer
    unit:
      ucum_code: By   
  staging_url:
    title: Staging Location
    description: URL for internal access to the data. May be temporary.
    range: uriorcurie
  release_url:
    title: Release Location
    description: URL for controlled or open access to the data.
    range: uriorcurie  
  drs_uri:
    title: DRS URI
    description: DRS location to access the data.
    range: uriorcurie
  hash:
    title: File Hash
    description: File hash information
    range: FileHash
  hash_type:
    title: File Hash Type
    description: The type of file hash, eg, md5
    range: EnumFileHashType
  hash_value:
    title: File Hash Value
    description: The value of the file hash
    range: string
  dataset_id:
    title: Dataset ID
    description: Unique identifier for a Dataset.
    range: Dataset
  data_collection_start:
    title: Data Collection Start
    description: The date that data collection started. May include only a year.
    #TODO: We could re-evaluate these as dates, but that may be too implementation specific
    range: string
  data_collection_end:
    title: Data Collection End
    description: The date that data collection started. May include only a year.
    #TODO: We could re-evaluate these as dates, but that may be too implementation specific
    range: string

enums:
  EnumProgram:
    title: Funding Programs
    description: Funding programs relevant to inform operations.
    permissible_values:
      include:
        title: INCLUDE
      kf:
        title: KF
      other:
        title: Other
  EnumResearchDomain:
    title: Research Domain
    description: Domains of Research used to find studies.
    #TODO: replace/add NIH internal categories from Huiqing
    permissible_values:
      behavior_and_behavior_mechanisms:
        title: Behavior and Behavior Mechanisms
        meaning: mesh:D001520
      congenital_heart_defects:
        title: Congenital Heart Defects
        meaning: mesh:D006330
      immune_system_diseases:
        title: Immune System Diseases
        meaning: mesh:D007154
      hematologic_diseases:
        title: Hematologic Diseases
        meaning: mesh:D006402
      neurodevelopment:
        title: Neurodevelopment
        meaning: mesh:D065886
      sleep_wake_disorders:
        title: Sleep Wake Disorders
        meaning: mesh:D012893
      all_co_occurring_conditions:
        title: All Co-occurring Conditions
        meaning: mesh:D013568
      physical_fitness:
        title: Physical Fitness
        meaning: mesh:D010809
      other:
        title: Other
  EnumParticipantLifespanStage:
    title: Participant Lifespan Stage
    description: Stages of life during which participants may be recruited.
    permissible_values:
      fetal:
        title: Fetal
        description: Before birth
      neonatal:
        title: Neonatal
        description: 0-28 days old
      pediatric:
        title: Pediatric
        description: Birth-17 years old
      adult:
        title: Adult
        description: 18+ years old
  EnumStudyDesign:
    title: Study Design
    description: Approaches for collecting data, investigating interventions, and/or analyzing data.
    #TODO: Add meanings
    permissible_values:
      case_control:
        title: Case-Control
      case_set:
        title: Case Set
      control_set:
        title: Control Set
      clinical_trial:
        title: Clinical Trial
      cross_sectional:
        title: Cross-Sectional
      family_twins_trios:
        title: Family/Twins/Trios
      interventional:
        title: Interventional
      longitudinal:
        title: Longitudinal
      trial_readiness_study:
        title: Trial Readiness Study
      tumor_vs_matched_normal:
        title: Tumor vs Matched Normal
  EnumClinicalDataSourceType:
    title: Clinical Data Source Type
    description: Approaches to ascertain clinical information about a participant.
    #TODO: Add meanings
    permissible_values:
      medical_record:
        title: Medical Record
        description: Data obtained directly from medical record
      investigator_assessment:
        title: Investigator Assessment
        description: Data obtained by examination, interview, etc. with investigator
      participant_or_caregiver_report:
        title: Participant or Caregiver Report
        description: Data obtained from survey, questionnaire, etc. filled out by participant or caregiver
      other:
        title: Other
        description: Data obtained from other source, such as tissue bank
      unknown:
        title: Unknown
  EnumDataCategory:
    title: Data Category
    description: Categories of data which may be collected about participants.
    #TODO: Add meanings
    permissible_values:
    #Should we have these two demo/clinical data categories?
      unharmonized_demographic_clinical_data:
        title: Unharmonized Demographic/Clinical Data
      harmonized_demographic_clinical_data:
        title: Harmonized Demographic/Clinical Data
      genomics:
        title: Genomics
      transcriptomics:
        title: Transcriptomics
      epigenomics:
        title: Epigenomics
      proteomics:
        title: Proteomics
      metabolomics:
        title: Metabolomics
      cognitive_behavioral:
        title: Cognitive/Behavioral
      immune_profiling:
        title: Immune Profiling
      imaging:
        title: Imaging
      microbiome:
        title: Microbiome
      #What is the difference between these two?
      fitness:
        title: Fitness
      physical_activity:
        title: Physical Activity
      other:
        title: Other
      sleep_study:
        title: Sleep Study
  EnumSubjectType:
    description: Types of Subject entities
    permissible_values: 
      participant: 
        description: Study participant with consent, assent, or waiver of consent.
      non_participant:
        description: An individual associated with a study who was not explictly consented, eg, the subject
          of a reported family history.
      cell_line: 
        description: Cell Line
      animal_model: 
        description: Animal model
      group:
        description: A group of individuals or entities.
      other:
        description: A different entity type- ideally this will be resolved!
  EnumDownSyndromeStatus:
    description: Down syndrome / chromosome 21 status
    permissible_values:
      d21:
        title: D21
        description: Disomy 21 (euploid)
        meaning: PATO:0001393
      t21:
        title: T21
        description: Trisomy 21 (Down syndrome)
        meaning: MONDO:0008608
  EnumSex:
    description: Subject Sex
    permissible_values:
      female:
        title: Female
        meaning: NCIT:C16576
      male:
        title: Male
        meaning: NCIT:C20197
      other:
        title: Other
        meaning: NCIT:C17649
      unknown:
        title: Unknown
        meaning: NCIT:C17998
  EnumRace:
    description: Participant Race
    permissible_values:
      american_indian_or_alaska_native:
        title: American Indian or Alaska Native
        meaning: NCIT:C41259
      asian:
        title: Asian
        meaning: NCIT:C41260
      black_or_african_american:
        title: Black or African American
        meaning: NCIT:C16352
      more_than_one_race:
        title: More than one race
        meaning: NCIT:C67109
      native_hawaiian_or_other_pacific_islander:
        title: Native Hawaiian or Other Pacific Islander
        meaning: NCIT:C41219
      other:
        title: Other
        meaning: NCIT:C17649
      white:
        title: White
        meaning: NCIT:C41261
      prefer_not_to_answer:
        title: Prefer not to answer
        meaning: NCIT:C132222
      unknown:
        title: Unknown
        meaning: NCIT:C17998
      east_asian:
        title: East Asian
        description: UK only; do not use for US data
        meaning: NCIT:C161419
      latin_american:
        title: Latin American
        description: UK only; do not use for US data
        meaning: NCIT:C126531
      middle_eastern_or_north_african:
        title: Middle Eastern or North African
        description: UK only; do not use for US data
        meaning: NCIT:C43866
      south_asian:
        title: South Asian
        description: UK only; do not use for US data
        meaning: NCIT:C41263
  EnumEthnicity:
    description: Participant ethnicity, specific to Hispanic or Latino.
    permissible_values:
      # asked_but_unknown:
      #   text: asked_but_unknown
      #   title: Asked but unknown
      hispanic_or_latino:
        title: Hispanic or Latino
        meaning: NCIT:C17459
      not_hispanic_or_latino:
        title: Not Hispanic or Latino
        meaning: NCIT:C41222
      #This should likely be rolled into a null flavor
      prefer_not_to_answer:
        title: Prefer not to answer
        meaning: NCIT:C132222
      unknown:
        title: Unknown
        meaning: NCIT:C17998
  EnumVitalStatus:
    description: Descriptions of a Subject's vital status
    is_a: EnumNull
    permissible_values:
      dead:
        title: Dead
        meaning: NCIT:C28554
      alive:
        title: Alive
        meaning: NCIT:C37987
  EnumNull:
    description: Base enumeration providing null options.
    permissible_values:
      unknown:
        title: Unknown
        meaning: NCIT:C17998
  EnumAssertionProvenance:
    description: Possible data sources for assertions.
    is_a: EnumNull
    permissible_values:
      medical_record:
        title: Medical Record
        description: Data obtained from a medical record
      investigator_assessment:
        title: Investigator Assessment
        description: Data obtained by examination, interview, etc. with investigator
      participant_or_caregiver_report:
        title: Participant or Caregiver Report
        description: Data obtained from survey, questionnaire, etc. filled out by participant or caregiver
      other:
        title: Other
        description: Data obtained from other source, such as tissue bank
  #TODO: Revisit these bindings / enumerations
  EnumAvailabilityStatus:
    description: Is the biospecimen available for use?
    permissible_values:
      available:
        title: Available 
        meaning: ig2_biospecimen_availability:available 
        description: Biospecimen is Available
      unavailable: 
        title: Unavailable 
        meaning: ig2_biospecimen_availability:unavailable 
        description: Biospecimen is Unavailable
  EnumSampleCollectionMethod:
    description: The approach used to collect the biospecimen. [LOINC](https://loinc.org) is recommended.
  EnumSite:
    description: The location of the specimen collection. [SNOMED Body Site](https://hl7.org/fhir/R4B/valueset-body-site.html) is recommended.
  EnumSpatialQualifiers:
    description: Any spatial/location qualifiers.
    enum_uri: http://hl7.org/fhir/us/mcode/ValueSet/mcode-body-location-qualifier-vs 
    reachable_from:
      source_ontology: bioregistry:snomedct
      source_nodes:
        - snomedct:106233006
        - snomedct:272424004 
        - snomedct:51440002
        - snomedct:399488007
        - snomedct:24028007
        - snomedct:7771000
      is_direct: false
      relationship_types:
        - rdfs:subClassOf
  EnumLaterality:
    description: Laterality information for the site
  EnumEDAMFormats:
    description: Data formats from the EDAM ontology.
    reachable_from:
      source_ontology: bioregistry:edam
      source_nodes:
        - edam:format_1915 #Format
      include_self: false
      is_direct: false
      relationship_types:
        - rdfs:subClassOf
  EnumEDAMDataTypes:
    description: Data types from the EDAM ontology.
    reachable_from:
      source_ontology: bioregistry:edam
      source_nodes:
        - edam:data_0006 #Data
      include_self: false
      is_direct: false
      relationship_types:
        - rdfs:subClassOf
  EnumFileHashType:
    description: Types of file hashes supported.
    permissible_values:
      md5:
        title: MD5
      etag: 
        title: ETag
      sha1:
        title: SHA-1