minimal_bdm_oeb_level2.yaml

$id: https://github.com/inab/benchmarking/minimal-json-schemas
$schema: 'https://json-schema.org/draft/2019-09/schema'
title: Minimal dataset concept (ELIXIR-EXCELERATE WP2 Benchmarking schemas)
if:
  type: array
then:
  $ref: '#/$defs/ArrayOfEntries'
else:
  properties:
    _id:
      title: The unique community-set id of the dataset
      type: string
    id:
      title: The unique community-set id of the challenge
      type: string
    type:
      type: string
      enum:
      - participant
      - assessment
      - aggregation
  if:
    required:
    - _id
  then:
    if:
      properties:
        type:
          const: participant
    then:
      $ref: '#/$defs/ParticipantEntry'
    else:
      $ref: '#/$defs/AssAggEntry'
  else:
    $ref: '#/$defs/ChallengeParticipants'
  
$defs:
  ArrayOfEntries:
    type: array
    minItems: 1
    items:
    - $ref: '#/$defs/ParticipantEntry'
    additionalItems:
      properties:
        _id:
          title: The unique community-set id of the dataset
          type: string
        id:
          title: The unique community-set id of the challenge
          type: string
      unevaluatedProperties: false
      if:
        required:
        - _id
      then:
        if:
          properties:
            type:
              const: participant
        then:
          $ref: '#/$defs/ParticipantEntry'
        else:
          $ref: '#/$defs/AssAggEntry'
      else:
        $ref: '#/$defs/ChallengeParticipants'
  
  ChallengeParticipants:
    type: object
    properties:
      id:
        title: The challenge acronym
        type: string
        minLength: 1
      participants:
        title: The list of participants
        type: array
        items:
          type: string
          minLength: 1
        uniqueItems: true
        minItems: 1
    additionalProperties: false
    required:
    - id
    - participants
    primary_key:
      name: challenges_key
      members:
      - id
  AssAggEntry:
    properties:
      type:
        title: Type of dataset, either assessment or aggregation
        type: string
        enum:
        - assessment
        - aggregation
    required:
    - type
    if:
      properties:
        type:
          const: assessment
    then:
      $ref: '#/$defs/AssessmentEntry'
    else:
      $ref: '#/$defs/AggregationEntry'
  AssessmentEntry:
    type: object
    properties:
      _id:
        title: The unique community-set id of the dataset
        type: string
      challenge_id:
        title: The challenge(s) where this dataset was generated or used
        type: string
        minLength: 1
# This foreign key is disabled because some QfO metrics workflows
# are producing spureous results
#        foreign_keys:
#        - members:
#          - .
#          schema_id: https://github.com/inab/benchmarking/minimal-json-schemas
#          refers_to: challenges_key
      community_id:
        title: The community where this dataset was generated or used
        type: string
        description: The name or OEB official id for the community which used or generated
          the dataset
        minLength: 1
#        foreign_keys:
#        - members:
#          - .
#          schema_id: https://www.elixir-europe.org/excelerate/WP2/json-schemas/1.0/Community
      metrics:
        title: The computed metrics stored as inline data
        type: object
        properties:
          metric_id:
            title: the OEB official id / name of the metric
            type: string
            minLength: 1
#            foreign_keys:
#            - members:
#              - .
#              schema_id: https://www.elixir-europe.org/excelerate/WP2/json-schemas/1.0/Metrics
        oneOf:
        - properties:
            values:
              title: the numerical value of the metric
              type: array
              minItems: 1
              oneOf:
              - items:
                  type: object
                  properties:
                    e:
                      title: the standard error / deviation for the metric (if exists)
                      type: number
                      minLength: 1
                    v:
                      title: the numerical value of the metric
                      type: number
                      minLength: 1
                  unevaluatedProperties: false
                  required:
                  - v
              - items:
                  type: number
          required:
          - values
        - properties:
            stderr:
              title: the standard error / deviation for the metric (if exists)
              type: number
              minLength: 1
            value:
              title: the numerical value of the metric
              type: number
              minLength: 1
          required:
          - value
        unevaluatedProperties: false
        required:
        - metric_id
      participant_id:
        title: The id / name of the tool which generated this dataset
        type: string
        minLength: 1
        join_keys:
        - members:
          - .
          schema_id: https://github.com/inab/benchmarking/minimal-json-schemas
          refers_to: participant_id_index
      type:
        title: Type of dataset.
        type: string
        const: assessment
    unevaluatedProperties: false
    required:
    - _id
    - community_id
    - challenge_id
    - type
    - participant_id
    primary_key:
    - _id
    unique:
    - challenge_id
    - community_id
    - participant_id
    - "metrics.metric_id"
  ParticipantEntry:
    type: object
    properties:
      _id:
        title: The unique community-set id of the dataset
        type: string
      challenge_id:
        title: The challenge(s) where this dataset was generated or used
        oneOf:
        - type: array
          items:
            type: string
            minLength: 1
            foreign_keys:
            - members:
              - .
              schema_id: https://github.com/inab/benchmarking/minimal-json-schemas
              refers_to: challenges_key
          minItems: 1
        - type: string
          minLength: 1
          foreign_keys:
          - members:
            - .
            schema_id: https://github.com/inab/benchmarking/minimal-json-schemas
            refers_to: challenges_key
      community_id:
        title: The community where this dataset was generated or used
        type: string
        description: The name or OEB official id for the community which used or generated
          the dataset
        minLength: 1
#        foreign_keys:
#        - members:
#          - .
#          schema_id: https://www.elixir-europe.org/excelerate/WP2/json-schemas/1.0/Community
      datalink:
        title: The data itself
        type: object
        properties:
          attrs:
            title: 'Attributes describing the datalink: whether it is inline (i.e.
              a data URI); a CURIE; whether it is compressed, or an archive (tar,
              zip, rar), a metalink (RFC 5854, RFC 6249, https://en.wikipedia.org/wiki/Metalink)
              or a standard sums (MD5, SHA1) file'
            type: array
            anyOf:
            - items:
                enum:
                - inline
                - compressed
                - archive
                - metalink
                - sums
                type: string
            - items:
                enum:
                - curie
                - compressed
                - archive
                - metalink
                type: string
            minItems: 1
            uniqueItems: true
          status:
            title: Last validation status (unreachable content, missing content(404),
              interrupted while validating, corrupted (checksums, not all the contents
              available, etc...), ok)
            type: string
            enum:
            - unreachable
            - missing
            - interrupted
            - corrupted
            - ok
          validation_date:
            title: Last validation date of the datalink
            type: string
            format: date-time
          uri:
            title: Where the dataset is available
            type: string
            format: uri
        additionalProperties: false
        required:
        - validation_date
        - status
      participant_id:
        title: The id / name of the tool which generated this dataset
        type: string
        minLength: 1
        index:
          name: participant_id_index
          members:
          - .
#        foreign_keys:
#        - members:
#          - .
#          schema_id: https://www.elixir-europe.org/excelerate/WP2/json-schemas/1.0/Tool
      type:
        title: Type of dataset.
        type: string
        const: participant
    additionalProperties: false
    required:
    - _id
    - community_id
    - challenge_id
    - type
    - datalink
    - participant_id
    primary_key:
    - _id
  AggregationEntry:
    type: object
    properties:
      _id:
        title: The unique community-set id of the dataset
        type: string
        minLength: 1
      challenge_ids:
        title: The challenge(s) where this dataset was generated or used
        type: array
        items:
          type: string
          minLength: 1
          foreign_keys:
          - members:
            - .
            schema_id: https://github.com/inab/benchmarking/minimal-json-schemas
            refers_to: challenges_key
        minItems: 1
      community_id:
        title: The community where this dataset was generated or used
        type: string
        description: The name or OEB official id for the community which used or generated
          the dataset
        minLength: 1
#        foreign_keys:
#        - members:
#          - .
#          schema_id: https://www.elixir-europe.org/excelerate/WP2/json-schemas/1.0/Community
      aggregation_metric_id:
        title: The OEB official id or label of the aggregation metrics to be associated to this aggregation dataset
        type: string
        minLength: 1
        description: >-
          The label of one of the metrics registered under the challenge declaration.
          This is needed to disambiguate the metrics to associate to the aggregation
          dataset when the challenge defines more than one.
      datalink:
        title: The aggregation data itself
        type: object
        properties:
          inline_data:
            title: The aggregation data is provided inline
            oneOf:
            - type: object
              properties:
                challenge_participants:
                  description: The metrics for each one of the participants
                  type: array
                  items:
                    type: object
                    properties:
                      participant_id:
                        type: string
                        minLength: 1
                      metric_x:
                        type: number
                      metric_y:
                        type: number
                      stderr_x:
                        type: number
                      stderr_y:
                        type: number
                    required:
                    - participant_id
                    - metric_x
                    - metric_y
                    unevaluatedProperties: false
                  unique:
                  - participant_id
                visualization:
                  type: object
                  properties:
                    type:
                      type: string
                      const: "2D-plot"
                    x_axis:
                      title: "x-axis metric label"
                      type: string
                      minLength: 1
                    y_axis:
                      title: "y-axis metric label"
                      type: string
                      minLength: 1
                    optimization:
                      title: "Chart representation optimization"
                      type: string
                      enum:
                      - "top-left"
                      - "top-right"
                      - "bottom-left"
                      - "bottom-right"
                      default: "top-right"
                  required:
                  - type
                  - x_axis
                  - y_axis
                  unevaluatedProperties: false
              required:
              - challenge_participants
              - visualization
              unevaluatedProperties: false
            - type: object
              properties:
                challenge_participants:
                  description: The metrics for each one of the participants
                  type: array
                  items:
                    type: object
                    properties:
                      participant_id:
                        type: string
                        minLength: 1
                      metric_value:
                        type: number
                    required:
                    - participant_id
                    - metric_value
                visualization:
                  type: object
                  properties:
                    type:
                      type: string
                      const: "bar-plot"
                    metric:
                      type: string
                      minLength: 1
                    optimization:
                      title: "Chart representation optimization"
                      type: string
                      enum:
                      - "left"
                      - "right"
                      - "minimize"
                      - "maximize"
                      default: "maximize"
                  required:
                  - type
                  - metric
                  unevaluatedProperties: false
              required:
              - challenge_participants
              - visualization
            - type: object
              properties:
                series_type:
                  type: string
                  const: aggregation-data-series
                challenge_participants:
                  description: The metrics data series for each one of the participants
                  type: array
                  items:
                    type: object
                    properties:
                      label:
                        type: string
                        minLength: 1
                      metric_id:
                        type: string
                        minLength: 1
                      values:
                        type: array
                        minItems: 1
                        oneOf:
                        - items:
                            type: object
                            properties:
                              v:
                                type: number
                              e:
                                type: number
                            unevaluatedProperties: false
                            required:
                            - v
                        - items:
                            type: number
                    join_keys:
                    - members:
                      - metric_id
                      schema_id: https://github.com/inab/benchmarking/minimal-json-schemas
                      refers_to: available_metrics_index
                    required:
                    - label
                    - metric_id
                    - values
                    unevaluatedProperties: false
                visualization:
                  type: object
                  properties:
                    type:
                      type: string
                      const: "box-plot"
                    available_metrics:
                      type: array
                      items:
                        type: string
                        minLength: 1
                      uniqueItems: true
                      minItems: 1
                    optimization:
                      title: Chart representation optimization (valid for bar-plot and 2D-plot)
                      type: string
                      enum:
                      - left
                      - right
                      - minimize
                      - maximize
                      - top-left
                      - top-right
                      - bottom-left
                      - bottom-right
                  required:
                  - type
                  - available_metrics
                  unevaluatedProperties: false
                labels:
                  title: Mapping labels to dataset assessment original ids. It can be an incomplete list.
                  type: array
                  items:
                    type: object
                    properties:
                      label:
                        title: Label representing the participant
                        type: string
                        minLength: 1
                      dataset_orig_id:
                        title: Participant dataset original id
                        type: string
                        minLength: 1
                    required:
                    - label
                    - dataset_orig_id
                    unevaluatedProperties: false
                  minItems: 1
              index:
                name: available_metrics_index
                members:
                - visualization.available_metrics
                limit_scope: true
              required:
              - series_type
              - challenge_participants
              - visualization
            unevaluatedProperties: false
      type:
        title: Type of dataset.
        type: string
        const: aggregation
    unevaluatedProperties: false
    required:
    - _id
    - challenge_ids
    - type
    - datalink
    unique:
    - _id
    - datalink.inline_data.visualization.x_axis
    - datalink.inline_data.visualization.y_axis