Merge pull request #2 from bmeg/fixtures

Fixtures setup

tests/fixtures/anvil/README.md

@@ -0,0 +1,4 @@
+## AnVIL
+
+### Download anvil_etl output
+gs://<google-bucket>/fhir/public

tests/fixtures/anvil/fhir/public/Public/1000G-high-coverage-2019/public/Organization.ndjson

@@ -0,0 +1,4 @@
+{"id":"anvil","identifier":[{"system":"https://anvil.terra.bio/program","value":"anvil"}],"resourceType":"Organization"}
+{"id":"Public","identifier":[{"system":"https://anvil.terra.bio/consortium","value":"Public"}],"partOf":{"reference":"Organization/anvil"},"resourceType":"Organization"}
+{"id":"public","identifier":[{"system":"https://anvil.terra.bio/#FHIR/data-store","value":"public"}],"partOf":{"reference":"Organization/Public"},"resourceType":"Organization"}
+{"id":"b44a3c8f-b197-52e2-9a3a-32d3d64c57c2","identifier":[{"system":"https://anvil.terra.bio/#workspaces/anvil-datastorage/","value":"1000G-high-coverage-2019"}],"partOf":{"reference":"Organization/public"},"resourceType":"Organization"}

tests/fixtures/anvil/fhir/public/Public/1000G-high-coverage-2019/public/Practitioner.ndjson

@@ -0,0 +1 @@
+{"id":"AnVIL-Team","resourceType":"Practitioner"}

tests/fixtures/anvil/fhir/public/Public/1000G-high-coverage-2019/public/PractitionerRole.ndjson

@@ -0,0 +1 @@
+{"id":"4a3b0c22-6314-57cf-88a7-852a7aff740a","identifier":[{"system":"https://anvil.terra.bio/#workspaces/anvil-datastorage/","value":"1000G-high-coverage-2019"}],"organization":{"reference":"Organization/b44a3c8f-b197-52e2-9a3a-32d3d64c57c2"},"practitioner":{"reference":"Practitioner/AnVIL-Team"},"resourceType":"PractitionerRole"}

tests/fixtures/anvil/fhir/public/Public/1000G-high-coverage-2019/public/ResearchStudy.ndjson

@@ -0,0 +1 @@
+{"id":"1000G-high-coverage-2019","identifier":[{"system":"https://anvil.terra.bio/#workspaces/anvil-datastorage/","value":"1000G-high-coverage-2019"},{"system":"https://anvil.terra.bio/#FHIR/data-store","value":"public"},{"system":"https://anvil.terra.bio/#consortium","value":"Public"}],"sponsor":{"reference":"Organization/b44a3c8f-b197-52e2-9a3a-32d3d64c57c2"},"status":"completed","resourceType":"ResearchStudy"}

tests/fixtures/anvil/fhir/public/Public/1000G-high-coverage-2019/public/ResearchStudyObservationSummary.ndjson

@@ -0,0 +1,120 @@
+{
+  "id": "c5b3fabc-35f6-5395-abbe-a01de4e2d64f",
+  "code": {
+    "coding": [
+      {
+        "code": "Summary",
+        "display": "Variable Summary",
+        "system": "https://nih-ncpi.github.io/ncpi-fhir-ig"
+      }
+    ]
+  },
+  "component": [
+    {
+      "code": {
+        "coding": [
+          {
+            "code": "SampleCount",
+            "display": "Number of Samples",
+            "system": "https://nih-ncpi.github.io/ncpi-fhir-ig"
+          }
+        ]
+      },
+      "valueInteger": 3202
+    },
+    {
+      "code": {
+        "coding": [
+          {
+            "code": "Participant",
+            "display": "Number of Participants",
+            "system": "https://nih-ncpi.github.io/ncpi-fhir-ig"
+          }
+        ]
+      },
+      "valueInteger": 3202
+    },
+    {
+      "code": {
+        "coding": [
+          {
+            "code": "StorageSize",
+            "display": "Size on Disk",
+            "system": "https://nih-ncpi.github.io/ncpi-fhir-ig"
+          }
+        ]
+      },
+      "valueQuantity": {
+        "code": "L",
+        "system": "http://unitsofmeasure.org",
+        "value": 90736987729057
+      }
+    },
+    {
+      "code": {
+        "coding": [
+          {
+            "code": "TerraWorkspace",
+            "display": "Terra Workspace Name",
+            "system": "https://nih-ncpi.github.io/ncpi-fhir-ig"
+          }
+        ]
+      },
+      "valueString": "1000G-high-coverage-2019"
+    },
+    {
+      "code": {
+        "coding": [
+          {
+            "code": "PrimaryDisease",
+            "display": "Primary Disease",
+            "system": "https://nih-ncpi.github.io/ncpi-fhir-ig"
+          }
+        ]
+      }
+    },
+    {
+      "code": {
+        "coding": [
+          {
+            "code": "StudyDesign",
+            "display": "Study Design",
+            "system": "https://nih-ncpi.github.io/ncpi-fhir-ig"
+          }
+        ]
+      },
+      "valueString": "Parent-Offspring Trios"
+    },
+    {
+      "code": {
+        "coding": [
+          {
+            "code": "DataType",
+            "display": "Data Type",
+            "system": "https://nih-ncpi.github.io/ncpi-fhir-ig"
+          }
+        ]
+      },
+      "valueString": "Whole Genome"
+    },
+    {
+      "code": {
+        "coding": [
+          {
+            "code": "ConsentCodes",
+            "display": "Consent Codes",
+            "system": "https://nih-ncpi.github.io/ncpi-fhir-ig"
+          }
+        ]
+      },
+      "valueString": "NRES"
+    }
+  ],
+  "focus": [
+    {
+      "reference": "ResearchStudy/1000G-high-coverage-2019"
+    }
+  ],
+  "status": "final",
+  "resourceType": "Observation"
+}

tests/fixtures/anvil/output/gen3/_definitions.yaml

@@ -0,0 +1,252 @@
+id: _definitions
+
+UUID:
+    term:
+        $ref: "_terms.yaml#/UUID"
+    type: string
+    pattern: "^[a-fA-F0-9]{8}-[a-fA-F0-9]{4}-[a-fA-F0-9]{4}-[a-fA-F0-9]{4}-[a-fA-F0-9]{12}$"
+
+parent_uuids:
+    type: array
+    minItems: 1
+    items:
+        $ref: "#/UUID"
+    uniqueItems: true
+
+foreign_key_project:
+    type: object
+    # Allow true here because we can have other unique keys defined on
+    # a target type
+    additionalProperties: true
+    #Can either use 'id' which are Gen3 IDs (UUID) or 'code'
+    #which is the user defined ID for project
+    properties:
+        id:
+            $ref: "#/UUID"
+        code:
+            type: string
+
+to_one_project:
+  anyOf:
+    - type: array
+      items:
+        $ref: "#/foreign_key_project"
+        minItems: 1
+        maxItems: 1
+    - $ref: "#/foreign_key_project"
+
+to_many_project:
+  anyOf:
+    - type: array
+      items:
+        $ref: "#/foreign_key_project"
+        minItems: 1
+    - $ref: "#/foreign_key_project"
+
+foreign_key:
+    type: object
+    # Allow true here because we can have other unique keys defined on
+    # a target type
+    additionalProperties: True
+    #Can either use 'id' which are GDC IDs (UUID) or 'submitter_id'
+    #which are user defined IDs ("submitter IDs in the backend")
+    properties:
+        id:
+            $ref: "#/UUID"
+        submitter_id:
+            type: string
+
+to_one:
+  anyOf:
+    - type: array
+      items:
+        $ref: "#/foreign_key"
+        minItems: 1
+        maxItems: 1
+    - $ref: "#/foreign_key"
+
+to_many:
+  anyOf:
+    - type: array
+      items:
+        $ref: "#/foreign_key"
+        minItems: 1
+    - $ref: "#/foreign_key"
+
+datetime:
+  oneOf:
+    - type: string
+      format: date-time
+    - type: 'null'
+  term:
+    $ref: "_terms.yaml#/datetime"
+
+file_name:
+    type: string
+    term:
+        $ref: "_terms.yaml#/file_name"
+
+file_size:
+    type: integer
+    term:
+        $ref: "_terms.yaml#/file_size"
+
+file_format:
+    type: string
+    term:
+        $ref: "_terms.yaml#/file_format"
+
+ga4gh_drs_uri:
+    type: string
+    term:
+        $ref: "_terms.yaml#/ga4gh_drs_uri"
+
+md5sum:
+    type: string
+    term:
+        $ref: "_terms.yaml#/md5sum"
+    pattern: "^[a-f0-9]{32}$"
+
+object_id:
+    type: string
+    description: "The GUID of the object in the index service."
+
+release_state:
+    description: "Release state of an entity."
+    default: unreleased
+    enum:
+      - unreleased
+      - released
+      - redacted
+
+data_bundle_state:
+    description: "State of a data bundle."
+    default: submitted
+    enum:
+        - submitted
+        - validated
+        - error
+        - released
+        - suppressed
+        - redacted
+
+data_file_error_type:
+    term:
+        $ref: "_terms.yaml#/data_file_error_type"
+    enum:
+        - file_size
+        - file_format
+        - md5sum
+
+state:
+    term:
+        $ref: "_terms.yaml#/state"
+    default: validated
+    downloadable:
+        - uploaded
+        - md5summed
+        - validating
+        - validated
+        - error
+        - invalid
+        - released
+    public:
+        - live
+    oneOf:
+        - enum: #This list covers legacy data
+            - uploading
+            - uploaded
+            - md5summing
+            - md5summed
+            - validating
+            - error
+            - invalid
+            - suppressed
+            - redacted
+            - live
+        - enum: #This list covers all future data 1/15/16
+            - validated
+            - submitted
+            - released
+
+file_state:
+    term:
+        $ref: "_terms.yaml#/file_state"
+    default: registered
+    enum:
+        - registered
+        - uploading
+        - uploaded
+        - validating
+        - validated
+        - submitted
+        - processing
+        - processed
+        - released
+        - error
+
+qc_metrics_state:
+    term:
+      $ref: "_terms.yaml#/qc_metric_state"
+    enum:
+        - FAIL
+        - PASS
+        - WARN
+
+project_id:
+    type: string
+    term:
+        $ref: "_terms.yaml#/project_id"
+
+data_file_properties:
+  $ref: "#/ubiquitous_properties"
+  file_name:
+    $ref: "#/file_name"
+  file_size:
+    $ref: "#/file_size"
+  file_format:
+    $ref: "#/file_format"  
+  md5sum:
+    $ref: "#/md5sum"
+  object_id:
+    $ref: "#/object_id"
+  file_state:
+    $ref: "#/file_state"
+  error_type:
+    $ref: "#/data_file_error_type"
+  ga4gh_drs_uri:
+    $ref: "#/ga4gh_drs_uri"
+
+workflow_properties:
+  $ref: "#/ubiquitous_properties"
+  workflow_link:
+    description: "Link to Github hash for the CWL workflow used."
+    type: string
+  workflow_version:
+    description: "Major version for a GDC workflow."
+    type: string
+  workflow_start_datetime:
+    $ref: "#/datetime"
+  workflow_end_datetime:
+    $ref: "#/datetime"
+
+ubiquitous_properties:
+  type:
+    type: string
+  id:
+    $ref: "#/UUID"
+    systemAlias: node_id
+  submitter_id:
+    type:
+      - string
+    description: >
+      A project-specific identifier for a node. This property is the calling card/nickname/alias for
+      a unit of submission. It can be used in place of the UUID for identifying or recalling a node.
+  state:
+    $ref: "#/state"
+  project_id:
+    $ref: "#/project_id"
+  created_datetime:
+    $ref: "#/datetime"
+  updated_datetime:
+    $ref: "#/datetime"

tests/fixtures/anvil/output/gen3/_settings.yaml

@@ -0,0 +1,6 @@
+# Global settings for the graph
+
+# Is the graph case centric, that we want
+# to create a link between all children to case
+# to expedite case filter on nodes
+enable_case_cache: false