[a r] Support for HCA bionetworks and atlases (#5519)

lambdas/service/app.py

@@ -228,7 +228,7 @@
         # changes and reset the minor version to zero. Otherwise, increment only
         # the minor version for backwards compatible changes. A backwards
         # compatible change is one that does not require updates to clients.
-        'version': '3.4'
+        'version': '3.5'
     },
     'tags': [
         {

src/azul/plugins/metadata/hca/__init__.py

@@ -203,7 +203,10 @@ def _field_mapping(self) -> MetadataPlugin._FieldMapping:
                     'project_title': 'projectTitle',
                     'publication_titles': 'publicationTitle',
                     'accessions': 'accessions',
-                    'estimated_cell_count': 'projectEstimatedCellCount'
+                    'estimated_cell_count': 'projectEstimatedCellCount',
+                    'is_tissue_atlas_project': 'isTissueAtlasProject',
+                    'tissue_atlas': 'tissueAtlas',
+                    'bionetwork_name': 'bionetworkName'
                 },
                 'sequencing_protocols': {
                     'instrument_manufacturer_model': 'instrumentManufacturerModel',
@@ -297,7 +300,10 @@ def facets(self) -> Sequence[str]:
             'projectDescription',
             'institution',
             'contactName',
-            'publicationTitle'
+            'publicationTitle',
+            'isTissueAtlasProject',
+            'tissueAtlas',
+            'bionetworkName'
         ]
 
     @property

src/azul/plugins/metadata/hca/indexer/aggregate.py

@@ -179,7 +179,8 @@ def _accumulator(self, field) -> Optional[Accumulator]:
                        'contact_names',
                        'contributors',
                        'publications',
-                       'accessions'):
+                       'accessions',
+                       'tissue_atlas'):
             return None
         elif field == 'estimated_cell_count':
             return MaxAccumulator()

src/azul/plugins/metadata/hca/indexer/transform.py

@@ -293,6 +293,8 @@ def api_schema(self) -> JSON:
 
 accession: Nested = Nested(namespace=null_str, accession=null_str)
 
+tissue_atlas: Nested = Nested(atlas=null_str, version=null_str)
+
 age_range = ClosedRange(pass_thru_float)
 
 
@@ -658,6 +660,12 @@ def _accession(self, p: api.Accession):
             'accession': p.accession
         }
 
+    def _tissue_atlas(self, b: api.Bionetwork):
+        return {
+            'atlas': b.hca_tissue_atlas,
+            'version': b.hca_tissue_atlas_version
+        }
+
     @classmethod
     def _project_types(cls) -> FieldTypes:
         return {
@@ -674,6 +682,9 @@ def _project_types(cls) -> FieldTypes:
             'supplementary_links': [null_str],
             '_type': null_str,
             'accessions': [accession],
+            'is_tissue_atlas_project': null_bool,
+            'tissue_atlas': [tissue_atlas],
+            'bionetwork_name': [null_str],
             'estimated_cell_count': null_int
         }
 
@@ -717,6 +728,10 @@ def _project(self, project: api.Project) -> MutableJSON:
             'supplementary_links': sorted(project.supplementary_links),
             '_type': 'project',
             'accessions': list(map(self._accession, project.accessions)),
+            'is_tissue_atlas_project': any(bionetwork.atlas_project
+                                           for bionetwork in project.bionetworks),
+            'tissue_atlas': list(map(self._tissue_atlas, project.bionetworks)),
+            'bionetwork_name': sorted(bionetwork.name for bionetwork in project.bionetworks),
             'estimated_cell_count': project.estimated_cell_count
         }
 

src/azul/plugins/metadata/hca/service/response.py

@@ -365,6 +365,9 @@ def make_projects(self, entry) -> MutableJSONs:
                 'projectShortname': project['project_short_name'],
                 'laboratory': sorted(set(project.get('laboratory', [None]))),
                 'estimatedCellCount': project['estimated_cell_count'],
+                'isTissueAtlasProject': project['is_tissue_atlas_project'],
+                'tissueAtlas': project.get('tissue_atlas'),
+                'bionetworkName': project['bionetwork_name']
             }
             if self.entity_type == 'projects':
                 translated_project['projectDescription'] = project.get('project_description', [])

src/humancellatlas/data/metadata/api.py

@@ -306,6 +306,15 @@ class Accession:
     accession: str
 
 
+@dataclass(eq=True, frozen=True)
+class Bionetwork:
+    name: str
+    atlas_project: Optional[bool] = None
+    hca_tissue_atlas: Optional[str] = None
+    hca_tissue_atlas_version: Optional[str] = None
+    schema_version: Optional[str] = None
+
+
 @dataclass(init=False)
 class Project(Entity):
     project_short_name: str
@@ -316,6 +325,7 @@ class Project(Entity):
     accessions: OrderedSet[Accession]
     supplementary_links: OrderedSet[str]
     estimated_cell_count: Optional[int]
+    bionetworks: OrderedSet[Bionetwork]
 
     def __init__(self,
                  json: JSON,
@@ -341,6 +351,9 @@ def __init__(self,
                 assert isinstance(value, list)
                 accessions.update(Accession(namespace=prefix, accession=v) for v in value)
         self.accessions = accessions
+        self.bionetworks = OrderedSet(Bionetwork(**bionetwork)
+                                      for bionetwork in content.get('hca_bionetworks', ())
+                                      if bionetwork)
 
     def _accessions(self, namespace: str) -> Set[str]:
         return {a.accession for a in self.accessions if a.namespace == namespace}

test/service/data/pfb_manifest.results.json

@@ -528,6 +528,9 @@
                 null
             ],
             "publications": [],
+            "is_tissue_atlas_project": [false],
+            "bionetwork_name": ["Immune"],
+            "tissue_atlas": [],
             "supplementary_links": [
                 null
             ]
@@ -864,6 +867,9 @@
                 "Single-Cell Analysis of Human Pancreas Reveals Transcriptional Signatures of Aging and Somatic Mutation Patterns."
             ],
             "publications": [],
+            "is_tissue_atlas_project": [false],
+            "bionetwork_name": [null],
+            "tissue_atlas": [],
             "supplementary_links": [
                 "https://www.ebi.ac.uk/gxa/sc/experiments/E-GEOD-81547/Results"
             ]
@@ -3273,6 +3279,9 @@
                 "The cellular immune response to COVID-19 deciphered by single cell multi-omics across three UK centres"
             ],
             "publications": [],
+            "is_tissue_atlas_project": [false],
+            "bionetwork_name": [null],
+            "tissue_atlas": [],
             "supplementary_links": [
                 null
             ]

test/service/data/pfb_manifest.schema.json

@@ -1936,6 +1936,51 @@
                                 "type": "array"
                             }
                         },
+                        {
+                            "name": "is_tissue_atlas_project",
+                            "namespace": "projects",
+                            "type": {
+                                "items": ["null", "boolean"],
+                                "type": "array"
+                            }
+                        },
+                        {
+                            "name": "tissue_atlas",
+                            "namespace": "projects",
+                            "type": {
+                                "items": {
+                                    "fields": [
+                                        {
+                                            "name": "atlas",
+                                               "namespace": "projects.tissue_atlas",
+                                               "type": {
+                                                   "items": ["null", "string"],
+                                                   "type": "array"
+                                               }
+                                        },
+                                        {
+                                            "name": "version",
+                                            "namespace": "projects.tissue_atlas",
+                                            "type": {
+                                                "items": ["null", "string"],
+                                                "type": "array"
+                                            }
+                                        }
+                                    ],
+                                   "name": "projects.tissue_atlas",
+                                   "type": "record"
+                                },
+                                "type": "array"
+                            }
+                        },
+                        {
+                            "name": "bionetwork_name",
+                            "namespace": "projects",
+                            "type": {
+                                "items": ["null", "string"],
+                                "type": "array"
+                            }
+                        },
                         {
                             "name": "estimated_cell_count",
                             "namespace": "projects",

test/service/test_index_projects.py

@@ -81,6 +81,9 @@ def assert_file_type_summaries(hit):
             'matrices',
             'contributedAnalyses',
             'accessions',
+            'tissueAtlas',
+            'isTissueAtlasProject',
+            'bionetworkName',
             'estimatedCellCount'
         }
         response_json = get_response_json()