Sync schemas

lang/R/inst/extdata/airr-schema.yaml

@@ -139,6 +139,63 @@ Attributes:
                             type: string
                             description: Ontology name for the top node term
 
+
+# A data set manifest groups a set of files that logically describe an AIRR data set
+# that spans a number of AIRR schema objects.
+
+DataSetManifest:
+    discriminator: AIRR
+    type: object
+        properties:
+            dataset_name:
+                type: string
+                description: A human readable name for the data set.
+            repository_url:
+                type: string
+                description: URL for the repository if the data set came from an ADC repository.
+            dataset:
+                type: array
+                description: An array that describes the type of data (and the relevant data files) in the data set.
+                items:
+                    type: object
+                    properties:
+                        data_type:
+                            type: string
+                            description:
+                                > A string that represents the type of the data that is being stored in the files.
+                                  If the data files store data from an AIRR Schema object, then the data_type should
+                                  be the schema object name (e.g. "Repertoire", "Rearrangement", "Clone", "Cell", "CellExpression",
+                                  "CellReactivity", "Receptor", etc.). If the type of files 
+                        data_files:
+                            type: array
+                            description: An array of file names that store the data of the specified data_type
+                            items:
+                                type: string
+
+
+# AIRR Manifest specification.
+#     An AIRR Manifest contains metadata for a group of accompanying files that are part of a set or coherent unit. 
+#     A manifest consists of an AIRR Info Object as well as an array of single data set manifests. Each data set
+#     manifest consists of tags for types of AIRR Schema Objects and a set of files that hold data in the AIRR format
+#     appropriate for that AIRR schema object.
+
+Manifest:
+    discriminator: AIRR
+    type: object
+    properties:
+        Info:
+            $ref: '#/InfoObject'
+            x-airr:
+                nullable: false
+        DataSets:
+            type: array
+            description: List of data sets that are described within this manifest
+            items:
+              $ref: '#/DataSetManifest'
+            x-airr:
+                nullable: false
+
+
 # AIRR Data File
 #
 # A JSON data file that holds Repertoire metadata, data processing
@@ -152,7 +209,7 @@ Attributes:
 
 # While the properties in an AIRR Data File are not required, if one is provided then
 # the value should not be null.
-
+        
 DataFile:
     discriminator: AIRR
     type: object

lang/python/airr/specs/airr-schema.yaml

@@ -139,6 +139,63 @@ Attributes:
                             type: string
                             description: Ontology name for the top node term
 
+
+# A data set manifest groups a set of files that logically describe an AIRR data set
+# that spans a number of AIRR schema objects.
+
+DataSetManifest:
+    discriminator: AIRR
+    type: object
+        properties:
+            dataset_name:
+                type: string
+                description: A human readable name for the data set.
+            repository_url:
+                type: string
+                description: URL for the repository if the data set came from an ADC repository.
+            dataset:
+                type: array
+                description: An array that describes the type of data (and the relevant data files) in the data set.
+                items:
+                    type: object
+                    properties:
+                        data_type:
+                            type: string
+                            description:
+                                > A string that represents the type of the data that is being stored in the files.
+                                  If the data files store data from an AIRR Schema object, then the data_type should
+                                  be the schema object name (e.g. "Repertoire", "Rearrangement", "Clone", "Cell", "CellExpression",
+                                  "CellReactivity", "Receptor", etc.). If the type of files 
+                        data_files:
+                            type: array
+                            description: An array of file names that store the data of the specified data_type
+                            items:
+                                type: string
+
+
+# AIRR Manifest specification.
+#     An AIRR Manifest contains metadata for a group of accompanying files that are part of a set or coherent unit. 
+#     A manifest consists of an AIRR Info Object as well as an array of single data set manifests. Each data set
+#     manifest consists of tags for types of AIRR Schema Objects and a set of files that hold data in the AIRR format
+#     appropriate for that AIRR schema object.
+
+Manifest:
+    discriminator: AIRR
+    type: object
+    properties:
+        Info:
+            $ref: '#/InfoObject'
+            x-airr:
+                nullable: false
+        DataSets:
+            type: array
+            description: List of data sets that are described within this manifest
+            items:
+              $ref: '#/DataSetManifest'
+            x-airr:
+                nullable: false
+
+
 # AIRR Data File
 #
 # A JSON data file that holds Repertoire metadata, data processing
@@ -152,7 +209,7 @@ Attributes:
 
 # While the properties in an AIRR Data File are not required, if one is provided then
 # the value should not be null.
-
+        
 DataFile:
     discriminator: AIRR
     type: object