feat: data submitters want collections to document their consortia

backend/layers/api/portal_api.py

@@ -30,6 +30,7 @@
     CollectionVersionException,
     DatasetInWrongStatusException,
     DatasetNotFoundException,
+    InvalidMetadataException,
     InvalidURIException,
     MaxFileSizeExceededException,
 )
@@ -217,7 +218,7 @@ def _collection_to_response(collection: CollectionVersionWithDatasets, access_ty
     is_tombstoned = collection.canonical_collection.tombstoned
     is_in_published_collection = collection.published_at is not None
 
-    return remove_none(
+    response = remove_none(
         {
             "access_type": access_type,
             "contact_email": collection.metadata.contact_email,
@@ -243,6 +244,10 @@ def _collection_to_response(collection: CollectionVersionWithDatasets, access_ty
         }
     )
 
+    # Always return consortia
+    response["consortia"] = collection.metadata.consortia
+    return response
+
 
 def lookup_collection(collection_id: str):
     """
@@ -337,11 +342,12 @@ def create_collection(body: dict, user: str):
         body["contact_name"],
         body["contact_email"],
         [_link_from_request(node) for node in body.get("links", [])],
+        body.get("consortia", []),
     )
 
     try:
         version = get_business_logic().create_collection(user, curator_name, metadata)
-    except CollectionCreationException as ex:
+    except (InvalidMetadataException, CollectionCreationException) as ex:
         raise InvalidParametersHTTPException(detail=ex.errors)
 
     return make_response(jsonify({"collection_id": version.collection_id.id}), 201)
@@ -424,9 +430,13 @@ def update_collection(collection_id: str, body: dict, token_info: dict):
         body.get("contact_name"),
         body.get("contact_email"),
         update_links,
+        body.get("consortia"),
     )
 
-    get_business_logic().update_collection_version(version.version_id, payload)
+    try:
+        get_business_logic().update_collection_version(version.version_id, payload)
+    except InvalidMetadataException as ex:
+        raise InvalidParametersHTTPException(detail=ex.errors)
 
     # Requires strong consistency w.r.t. the operation above - if not available, the update needs
     # to be done in memory

backend/layers/business/business.py

@@ -102,6 +102,8 @@ def create_collection(
         """
 
         errors = []
+        # Check metadata is valid
+        collection_metadata.strip_fields()
         validation.verify_collection_metadata(collection_metadata, errors)
 
         # TODO: Maybe switch link.type to be an enum
@@ -115,7 +117,6 @@ def create_collection(
         if errors:
             raise CollectionCreationException(errors)
 
-        collection_metadata.strip_fields()
         created_version = self.database_provider.create_canonical_collection(owner, curator_name, collection_metadata)
 
         # TODO: can collapse with `create_canonical_collection`
@@ -210,6 +211,8 @@ def update_collection_version(self, version_id: CollectionVersionId, body: Colle
         # TODO: link.type should DEFINITELY move to an enum. pylance will help with the refactor
 
         errors = []
+
+        # Check metadata
         validation.verify_collection_metadata_update(body, errors)
 
         current_version = self.get_collection_version(version_id)

backend/layers/business/entities.py

@@ -34,3 +34,4 @@ class CollectionMetadataUpdate:
     contact_name: Optional[str]
     contact_email: Optional[str]
     links: Optional[List[Link]]
+    consortia: Optional[List[str]]

backend/layers/business/exceptions.py

@@ -41,6 +41,12 @@ def __init__(self, errors: Optional[List[str]] = None) -> None:
         super().__init__()
 
 
+class InvalidMetadataException(BusinessException):
+    def __init__(self, errors: Optional[List[str]] = None) -> None:
+        self.errors: Optional[List[str]] = errors
+        super().__init__()
+
+
 class CollectionVersionException(BusinessException):
     pass
 

backend/layers/common/entities.py

@@ -1,4 +1,4 @@
-from dataclasses import dataclass
+from dataclasses import dataclass, field
 from datetime import datetime
 from enum import Enum
 from typing import List, Optional
@@ -211,6 +211,7 @@ class CollectionMetadata:
     contact_name: str
     contact_email: str
     links: List[Link]
+    consortia: List[str] = field(default_factory=list)
 
     def strip_fields(self):
         self.name = self.name.strip()
@@ -219,6 +220,8 @@ def strip_fields(self):
         self.contact_email = self.contact_email.strip()
         for link in self.links:
             link.strip_fields()
+        if self.consortia is not None:
+            self.consortia = [consortium.strip() for consortium in self.consortia]
 
 
 @dataclass

backend/layers/common/validation.py

@@ -3,12 +3,31 @@
 
 from urllib.parse import urlparse
 from backend.layers.business.entities import CollectionMetadataUpdate
+from backend.layers.business.exceptions import InvalidMetadataException
 
 from backend.layers.common.entities import CollectionMetadata, Link
 from backend.layers.common.regex import CONTROL_CHARS, EMAIL_REGEX
 
 control_char_re = re.compile(CONTROL_CHARS)
 
+valid_consortia = {
+    "Allen Institute for Brain Science",
+    "BRAIN Initiative",
+    "CZ Biohub",
+    "CZI Neurodegeneration Challenge Network",
+    "CZI Single-Cell Biology",
+    "European Union’s Horizon 2020",
+    "GenitoUrinary Development Molecular Anatomy Project (GUDMAP)",
+    "Gut Cell Atlas",
+    "Human BioMolecular Atlas Program (HuBMAP)",
+    "Human Pancreas Analysis Program (HPAP)",
+    "Human Tumor Atlas Network (HTAN)",
+    "Kidney Precision Medicine Project (KPMP)",
+    "LungMAP",
+    "SEA-AD",
+    "Wellcome HCA Strategic Science Support",
+}
+
 
 def _verify_collection_metadata_fields(
     metadata: Union[CollectionMetadata, CollectionMetadataUpdate], check_existence: bool, errors: list
@@ -33,6 +52,13 @@ def check(key):
         else:
             return value
 
+    def verify_collection_consortia(metadata: Union[CollectionMetadata, CollectionMetadataUpdate], errors: list):
+        consortia = getattr(metadata, "consortia")
+        if consortia:
+            for consortium in consortia:
+                if consortium not in valid_consortia:
+                    errors.append({"name": "consortia", "reason": "Invalid consortia."})
+
     contact_email = check("contact_email")
     if contact_email:
         result = EMAIL_REGEX.match(contact_email)
@@ -43,6 +69,8 @@ def check(key):
     check("name")
     check("contact_name")
 
+    verify_collection_consortia(metadata, errors)
+
 
 def verify_collection_links(links: List[Link], errors: list) -> None:
     def _error_message(i: int, _url: str) -> dict:
@@ -60,6 +88,8 @@ def _error_message(i: int, _url: str) -> dict:
 
 def verify_collection_metadata_update(metadata: CollectionMetadataUpdate, errors: list) -> None:
     _verify_collection_metadata_fields(metadata, check_existence=False, errors=errors)
+    if errors:
+        raise InvalidMetadataException(errors=errors)
     if metadata.links is not None:
         verify_collection_links(metadata.links, errors)
 
@@ -69,4 +99,6 @@ def verify_collection_metadata(metadata: CollectionMetadata, errors: list) -> No
     Verify if `CollectionMetadata` is well defined. Since `CollectionMetadataCreate
     """
     _verify_collection_metadata_fields(metadata, check_existence=True, errors=errors)
+    if errors:
+        raise InvalidMetadataException(errors=errors)
     verify_collection_links(metadata.links, errors)

backend/portal/api/app/portal-api.yml

@@ -103,6 +103,8 @@ paths:
                   description: name of the curator of the collection.
                 links:
                   $ref: "#/components/schemas/links"
+                consortia:
+                  $ref: "#/components/schemas/consortia"
       responses:
         "201":
           description: A new collection has been created.
@@ -208,6 +210,8 @@ paths:
                   description: name of the curator of the collection
                 links:
                   $ref: "#/components/schemas/links"
+                consortia:
+                  $ref: "#/components/schemas/consortia"
       responses:
         "200":
           description: OK
@@ -445,6 +449,8 @@ paths:
                       type: number
                     revised_at:
                       type: number
+                    consortia:
+                      $ref: "#/components/schemas/consortia"
         "400":
           $ref: "#/components/responses/400"
 
@@ -977,6 +983,11 @@ components:
           link_type:
             type: string
             enum: [PROTOCOL, RAW_DATA, DOI, LAB_WEBSITE, OTHER, DATA_SOURCE]
+    consortia:
+      type: array
+      items:
+        type: string
+        description: The broader research groups which contributed to this collection of datasets.
     collection:
       type: object
       required:
@@ -986,6 +997,7 @@ components:
         - visibility
         - links
         - datasets
+        - consortia
         - created_at
         - updated_at
         - data_submission_policy_version
@@ -1007,6 +1019,8 @@ components:
           $ref: "#/components/schemas/visibility"
         links:
           $ref: "#/components/schemas/links"
+        consortia:
+          $ref: "#/components/schemas/consortia"
         contact_name:
           type: string
         contact_email:

backend/portal/api/curation/curation-api.yml

@@ -586,6 +586,8 @@ components:
             The CELLxGENE Discover URL for the Collection. This points to the canonical link unless it's a revision, 
             in which case it points to the unpublished revision version.
           type: string
+        consortia:
+          $ref: "#/components/schemas/consortia"
         contact_email:
           $ref: "#/components/schemas/contact_email"
         contact_name:
@@ -627,6 +629,11 @@ components:
     collection_url:
       description: The CELLxGENE Discover URL for the Collection.
       type: string
+    consortia:
+      type: array
+      items:
+        type: string
+        description: The broader research groups which contributed to this collection of datasets.
     contact_email:
       description: The email of contact person for the Collection. Example email john.smith@email.com
       type: string
@@ -681,6 +688,8 @@ components:
       properties:
         collection_url:
           $ref: "#/components/schemas/collection_url"
+        consortia:
+          $ref: "#/components/schemas/consortia"
         contact_email:
           $ref: "#/components/schemas/contact_email"
         contact_name:
@@ -721,6 +730,8 @@ components:
       type: object
     collection_form_metadata:
       properties:
+        consortia:
+          $ref: "#/components/schemas/consortia"
         contact_email:
           $ref: "#/components/schemas/contact_email"
         contact_name:

backend/portal/api/curation/v1/curation/collections/actions.py

@@ -5,7 +5,7 @@
 from backend.layers.api.providers import get_business_logic
 from backend.layers.auth.user_info import UserInfo
 from backend.layers.business.entities import CollectionQueryFilter
-from backend.layers.business.exceptions import CollectionCreationException
+from backend.layers.business.exceptions import CollectionCreationException, InvalidMetadataException
 from backend.layers.common import doi
 from backend.layers.common.entities import CollectionMetadata, Link
 from backend.portal.api.curation.v1.curation.collections.common import reshape_for_curation_api
@@ -58,10 +58,19 @@ def post(body: dict, user: str):
 
     # Build CollectionMetadata object
     links = [Link(link.get("link_name"), link["link_type"], link["link_url"]) for link in body.get("links", [])]
-    metadata = CollectionMetadata(body["name"], body["description"], body["contact_name"], body["contact_email"], links)
+    metadata = CollectionMetadata(
+        body["name"],
+        body["description"],
+        body["contact_name"],
+        body["contact_email"],
+        links,
+        body.get("consortia", []),
+    )
 
     try:
         version = get_business_logic().create_collection(user, "", metadata)
+    except InvalidMetadataException as ex:
+        errors.extend(ex.errors)
     except CollectionCreationException as ex:
         errors.extend(ex.errors)
     if errors:

backend/portal/api/curation/v1/curation/collections/collection_id/actions.py

@@ -10,7 +10,7 @@
 from backend.layers.api.providers import get_business_logic
 from backend.layers.auth.user_info import UserInfo
 from backend.layers.business.entities import CollectionMetadataUpdate
-from backend.layers.business.exceptions import CollectionUpdateException
+from backend.layers.business.exceptions import CollectionUpdateException, InvalidMetadataException
 from backend.layers.common import doi
 from backend.layers.common.entities import Link
 from backend.portal.api.curation.v1.curation.collections.common import (
@@ -81,11 +81,14 @@ def _link_from_request(body: dict):
         body.get("contact_name"),
         body.get("contact_email"),
         update_links,
+        body.get("consortia", []),
     )
 
     # Update the collection
     try:
         get_business_logic().update_collection_version(collection_version.version_id, collection_metadata)
+    except InvalidMetadataException as ex:
+        raise InvalidParametersHTTPException(ext=dict(invalid_parameters=ex.errors))
     except CollectionUpdateException as ex:
         errors.extend(ex.errors)
     if errors:

backend/portal/api/curation/v1/curation/collections/common.py

@@ -108,6 +108,7 @@ def reshape_for_curation_api(
         revised_at = None
     response = dict(
         collection_url=collection_url,
+        consortia=collection_version.metadata.consortia,
         contact_email=collection_version.metadata.contact_email,
         contact_name=collection_version.metadata.contact_name,
         created_at=collection_version.created_at,
@@ -215,6 +216,7 @@ class EntityColumns:
         "links",
         "datasets",
         "revising_in",
+        "consortia",
     ]
 
     link_cols = [

frontend/src/common/entities.ts

@@ -75,6 +75,7 @@ export interface Author {
 
 export interface Collection {
   access_type: ACCESS_TYPE;
+  consortia: string[];
   contact_email: string;
   contact_name: string;
   description: string;

frontend/src/components/CreateCollectionModal/components/Content/common/constants.ts

@@ -1,3 +1,21 @@
+export enum CONSORTIA {
+  ALLEN_INSTITUTE_FOR_BRAIN_SCIENCE = "Allen Institute for Brain Science",
+  BRAIN_INITIATIVE = "BRAIN Initiative",
+  CZ_BIOHUB = "CZ Biohub",
+  CZI_NDCN = "CZI Neurodegeneration Challenge Network",
+  CZI_SCB = "CZI Single-Cell Biology",
+  EU_HORIZON_2020 = "European Union’s Horizon 2020",
+  GUDMAP = "GenitoUrinary Development Molecular Anatomy Project (GUDMAP)",
+  GUT_CELL_ATLAS = "Gut Cell Atlas",
+  HUBMAP = "Human BioMolecular Atlas Program (HuBMAP)",
+  HPAP = "Human Pancreas Analysis Program (HPAP)",
+  HTAN = "Human Tumor Atlas Network (HTAN)",
+  KPMP = "Kidney Precision Medicine Project (KPMP)",
+  LUNG_MAP = "LungMAP",
+  SEA_AD = "SEA-AD",
+  WELCOME_HCA_STRATEGIC_SCIENCE_SUPPORT = "Wellcome HCA Strategic Science Support",
+}
+
 export const DEBOUNCE_TIME_MS = 100;
 
 /**