Merge pull request #40 from uc-cdis/test

test

Author: cgmeyer

gdcdictionary/schemas/acknowledgement.yaml

@@ -3,11 +3,12 @@ $schema: "http://json-schema.org/draft-04/schema#"
 id: "acknowledgement"
 title: Acknowledgement
 type: object
+nodeTerms: null
 namespace: https://nci-crdc-demo.datacommons.io/
 category: administrative
 program: '*'
 project: '*'
-description: "Acknowledgement of an individual involved in a project."
+description: Acknowledgement of an individual involved in a project.
 additionalProperties: false
 submittable: true
 validators: null
@@ -33,14 +34,15 @@ required:
   - projects
 
 uniqueKeys:
-  - [ id ]
-  - [ project_id, submitter_id ]
+  - [id]
+  - [project_id, submitter_id]
 
 properties:
+
   $ref: "_definitions.yaml#/ubiquitous_properties"
 
   acknowledgee:
-    description: "The indvidiual or group being acknowledged by the project."
+    description: The indvidiual or group being acknowledged by the project.
     type: string
 
   projects:

gdcdictionary/schemas/aggregated_genotyping_array.yaml

@@ -3,21 +3,23 @@ $schema: "http://json-schema.org/draft-04/schema#"
 id: "aggregated_genotyping_array"
 title: Aggregated Genotyping array
 type: object
+nodeTerms: null
 namespace: https://nci-crdc-demo.datacommons.io/
 category: data_file
 program: '*'
 project: '*'
-description: "DNA array data files aggregating genome-wide SNP information from several subjects (cohort)."
+description: DNA array data files aggregating genome-wide SNP information from several
+  subjects (cohort).
 additionalProperties: false
 submittable: true
 validators: null
 
 systemProperties:
   - id
   - project_id
+  - state
   - created_datetime
   - updated_datetime
-  - state
   - file_state
   - error_type
 
@@ -51,12 +53,15 @@ required:
   - data_category
   - data_type
   - data_format
+  - file_size
+  - md5sum
 
 uniqueKeys:
-  - [ id ]
-  - [ project_id, submitter_id ]
+  - [id]
+  - [project_id, submitter_id]
 
 properties:
+
   $ref: "_definitions.yaml#/data_file_properties"
 
   data_category:
@@ -65,20 +70,20 @@ properties:
     enum:
       - SNP Array Data
 
-  data_type:
-    term:
-      $ref: "_terms.yaml#/data_type"
-    enum:
-      - Aggregated SNP Alleles
-      - Individual SNP Alleles
-
   data_format:
     term:
       $ref: "_terms.yaml#/data_format"
     enum:
+      - IND
       - PED
       - VCF
-      - IND
+
+  data_type:
+    term:
+      $ref: "_terms.yaml#/data_type"
+    enum:
+      - Aggregated SNP Alleles
+      - Individual SNP Alleles
 
   experimental_strategy:
     term:
@@ -90,14 +95,16 @@ properties:
     term:
       $ref: "_terms.yaml#/platform"
     enum:
+      - Affymetrix Genome-Wide Human SNP Array 6.0
       - Illumina Infinium HumanExome BeadChip
-      - Illumina OMNI 5M SNP Array
       - Illumina OMNI 2.5M SNP Array
-      - Affymetrix Genome-Wide Human SNP Array 6.0
+      - Illumina OMNI 5M SNP Array
+
+  core_metadata_collections:
+    $ref: "_definitions.yaml#/to_one"
 
   genotyping_arrays:
     $ref: "_definitions.yaml#/to_many"
+
   aliquots:
     $ref: "_definitions.yaml#/to_many"
-  core_metadata_collections:
-    $ref: "_definitions.yaml#/to_many"

gdcdictionary/schemas/aligned_reads.yaml

@@ -1,24 +1,24 @@
-chema: "http://json-schema.org/draft-04/schema#"
+$schema: "http://json-schema.org/draft-04/schema#"
 
 id: "aligned_reads"
 title: Aligned Reads
 type: object
+nodeTerms: null
 namespace: https://nci-crdc-demo.datacommons.io/
 category: data_file
 program: '*'
 project: '*'
-description: >
-  Data file containing aligned reads that are generated internally.
+description: Data file containing aligned reads that are generated internally.
 additionalProperties: false
 submittable: true
 validators: null
 
 systemProperties:
   - id
   - project_id
+  - state
   - created_datetime
   - updated_datetime
-  - state
   - file_state
   - error_type
 
@@ -70,24 +70,20 @@ required:
   - platform
 
 uniqueKeys:
-  - [ id ]
-  - [ project_id, submitter_id ]
+  - [id]
+  - [project_id, submitter_id]
 
 properties:
+
   $ref: "_definitions.yaml#/data_file_properties"
 
   data_category:
     term:
       $ref: "_terms.yaml#/data_category"
     enum:
+      - Raw Sequencing Data
       - Sequencing Data
       - Sequencing Reads
-      - Raw Sequencing Data
-
-  data_type:
-    term:
-      $ref: "_terms.yaml#/data_type"
-    enum: [ "Aligned Reads" ]
 
   data_format:
     term:
@@ -96,37 +92,47 @@ properties:
       - BAM
       - CRAM
 
+  data_type:
+    term:
+      $ref: "_terms.yaml#/data_type"
+    enum:
+      - Aligned Reads
+
   experimental_strategy:
     term:
       $ref: "_terms.yaml#/experimental_strategy"
     enum:
-      - WGS
-      - WXS
       - Low Pass WGS
-      - Validation
       - RNA-Seq
-      - miRNA-Seq
       - Total RNA-Seq
+      - Validation
+      - WGS
+      - WXS
+      - miRNA-Seq
 
   platform:
     term:
       $ref: "_terms.yaml#/platform"
     enum:
-      - "Illumina"
-      - "SOLiD"
-      - "LS454"
-      - "Ion Torrent"
-      - "Complete Genomics"
-      - "PacBio"
-      - "Other"
+      - Complete Genomics
+      - Illumina
+      - Ion Torrent
+      - LS454
+      - Other
+      - PacBio
+      - SOLiD
+
+  core_metadata_collections:
+    $ref: "_definitions.yaml#/to_one"
 
   alignment_cocleaning_workflows:
     $ref: "_definitions.yaml#/to_one"
+
   alignment_workflows:
     $ref: "_definitions.yaml#/to_one"
+
   submitted_unaligned_reads_files:
     $ref: "_definitions.yaml#/to_many"
+
   submitted_aligned_reads_files:
     $ref: "_definitions.yaml#/to_one"
-  core_metadata_collections:
-    $ref: "_definitions.yaml#/to_many"

gdcdictionary/schemas/aligned_reads_index.yaml

@@ -3,21 +3,22 @@ $schema: "http://json-schema.org/draft-04/schema#"
 id: "aligned_reads_index"
 title: Aligned Reads Index
 type: object
+nodeTerms: null
 namespace: https://nci-crdc-demo.datacommons.io/
 category: index_file
 program: '*'
 project: '*'
-description: "Data file containing the index for a set of aligned reads."
+description: Data file containing the index for a set of aligned reads.
 additionalProperties: false
 submittable: true
 validators: null
 
 systemProperties:
   - id
   - project_id
+  - state
   - created_datetime
   - updated_datetime
-  - state
   - file_state
   - error_type
 
@@ -49,29 +50,32 @@ required:
   - data_format
 
 uniqueKeys:
-  - [ id ]
-  - [ project_id, submitter_id ]
+  - [id]
+  - [project_id, submitter_id]
 
 properties:
+
   $ref: "_definitions.yaml#/data_file_properties"
 
   data_category:
     term:
       $ref: "_terms.yaml#/data_category"
     enum:
+      - Raw Sequencing Data
       - Sequencing Data
       - Sequencing Reads
-      - Raw Sequencing Data
-
-  data_type:
-    term:
-      $ref: "_terms.yaml#/data_type"
-    enum: [ "Aligned Reads Index" ]
 
   data_format:
     term:
       $ref: "_terms.yaml#/data_format"
-    enum: [ "BAI" ]
+    enum:
+      - BAI
+
+  data_type:
+    term:
+      $ref: "_terms.yaml#/data_type"
+    enum:
+      - Aligned Reads Index
 
   submitted_aligned_reads_files:
     $ref: "_definitions.yaml#/to_one"

gdcdictionary/schemas/alignment_cocleaning_workflow.yaml

@@ -3,23 +3,23 @@ $schema: "http://json-schema.org/draft-04/schema#"
 id: "alignment_cocleaning_workflow"
 title: Alignment Cocleaning Workflow
 type: object
+nodeTerms: null
 namespace: https://nci-crdc-demo.datacommons.io/
 category: analysis
 program: '*'
 project: '*'
-description: >
-  Metadata for the alignment and cocleaning pipeline used to align reads in the GDC harmonization
-  pipelines.
+description: Metadata for the alignment and cocleaning pipeline used to align reads
+  in the GDC harmonization pipelines.
 additionalProperties: false
 submittable: true
 validators: null
 
 systemProperties:
   - id
   - project_id
+  - state
   - created_datetime
   - updated_datetime
-  - state
 
 links:
   - exclusive: true
@@ -45,16 +45,18 @@ required:
   - workflow_type
 
 uniqueKeys:
-  - [ id ]
-  - [ project_id, submitter_id ]
+  - [id]
+  - [project_id, submitter_id]
 
 properties:
+
   $ref: "_definitions.yaml#/workflow_properties"
 
   workflow_type:
     term:
       $ref: "_terms.yaml#/workflow_type"
-    enum: [ "BWA with Mark Duplicates and Cocleaning" ]
+    enum:
+      - BWA with Mark Duplicates and Cocleaning
 
   submitted_aligned_reads_files:
     $ref: "_definitions.yaml#/to_many"