Index all AnVIL v4 schema columns (#4617)

DataBiosphere · Dec 6, 2022 · a49cd09 · a49cd09
1 parent 8157dd3
commit a49cd09
Show file tree

Hide file tree

Showing 4 changed files with 33 additions and 4 deletions.
diff --git a/src/azul/plugins/metadata/anvil/__init__.py b/src/azul/plugins/metadata/anvil/__init__.py
@@ -107,9 +107,9 @@ def _field_mapping(self) -> MetadataPlugin._FieldMapping:
                         'activity_table',
                         'activity_type',
                         'assay_category',
+                        'assay_type',
                         'data_modality',
-                        'document_id',
-                        'source_datarepo_row_ids',
+                        'reference_assembly',
                         # Not in schema
                         'date_created',
                     ]
@@ -119,6 +119,7 @@ def _field_mapping(self) -> MetadataPlugin._FieldMapping:
                         *common_fields,
                         'biosample_id',
                         'anatomical_site',
+                        'apriori_cell_type',
                         'biosample_type',
                         'disease',
                         'donor_age_at_collection_unit',
@@ -131,8 +132,11 @@ def _field_mapping(self) -> MetadataPlugin._FieldMapping:
                         'dataset_id',
                         'consent_group',
                         'data_use_permission',
+                        'owner',
+                        'principal_investigator',
                         'registered_identifier',
                         'title',
+                        'data_modality',
                     ]
                 },
                 'donors': {
@@ -142,6 +146,7 @@ def _field_mapping(self) -> MetadataPlugin._FieldMapping:
                         'organism_type',
                         'phenotypic_sex',
                         'reported_ethnicity',
+                        'genetic_ancestry',
                     ]
                 },
                 'files': {
@@ -151,7 +156,9 @@ def _field_mapping(self) -> MetadataPlugin._FieldMapping:
                             'file_id',
                             'data_modality',
                             'file_format',
+                            'file_md5sum',
                             'reference_assembly',
+                            'file_name',
                             # Not in schema
                             'crc32',
                             'sha256',
@@ -180,6 +187,7 @@ def facets(self) -> Sequence[str]:
         return [
             'activities.activity_type',
             'activities.assay_category',
+            'activities.assay_type',
             'activities.data_modality',
             'biosamples.anatomical_site',
             'biosamples.biosample_type',

diff --git a/src/azul/plugins/metadata/anvil/indexer/transform.py b/src/azul/plugins/metadata/anvil/indexer/transform.py
@@ -196,7 +196,9 @@ def _activity_types(cls) -> FieldTypes:
             'activity_table': null_str,
             'activity_type': null_str,
             'assay_category': null_str,
+            'assay_type': null_str,
             'data_modality': null_str,
+            'reference_assembly': [null_str],
             # Not in schema
             'date_created': null_datetime,
         }
@@ -207,6 +209,7 @@ def _biosample_types(cls) -> FieldTypes:
             **cls._entity_types(),
             'biosample_id': null_str,
             'anatomical_site': null_str,
+            'apriori_cell_type': [null_str],
             'biosample_type': null_str,
             'disease': null_str,
             'donor_age_at_collection_unit': null_str,
@@ -220,8 +223,11 @@ def _dataset_types(cls) -> FieldTypes:
             'dataset_id': null_str,
             'consent_group': [null_str],
             'data_use_permission': [null_str],
+            'owner': [null_str],
+            'principal_investigator': [null_str],
             'registered_identifier': [null_str],
             'title': null_str,
+            'data_modality': [null_str],
         }
 
     @classmethod
@@ -232,6 +238,7 @@ def _donor_types(cls) -> FieldTypes:
             'organism_type': null_str,
             'phenotypic_sex': null_str,
             'reported_ethnicity': null_str,
+            'genetic_ancestry': [null_str],
         }
 
     @classmethod
@@ -242,7 +249,9 @@ def _file_types(cls) -> FieldTypes:
             'data_modality': [null_str],
             'file_format': null_str,
             'file_size': null_int,
+            'file_md5sum': null_str,
             'reference_assembly': [null_str],
+            'file_name': null_str,
             # Not in schema
             'version': null_str,
             'uuid': null_str,

diff --git a/src/azul/plugins/metadata/anvil/service/response.py b/src/azul/plugins/metadata/anvil/service/response.py
@@ -182,6 +182,7 @@ def _non_pivotal_entity(self,
         'activities': {
             'activity_type',
             'assay_category',
+            'assay_type',
             'data_modality'
         },
         'biosamples': {
@@ -198,7 +199,8 @@ def _non_pivotal_entity(self,
         'donors': {
             'organism_type',
             'phenotypic_sex',
-            'reported_ethnicity'
+            'reported_ethnicity',
+            'genetic_ancestry'
         },
         'files': {
             'count',

diff --git a/src/azul/plugins/repository/tdr_anvil/__init__.py b/src/azul/plugins/repository/tdr_anvil/__init__.py
@@ -496,6 +496,7 @@ def convert_column(value):
         'biosample': {
             'biosample_id',
             'anatomical_site',
+            'apriori_cell_type',
             'biosample_type',
             'disease',
             'donor_age_at_collection_unit',
@@ -506,21 +507,27 @@ def convert_column(value):
             'dataset_id',
             'consent_group',
             'data_use_permission',
+            'owner',
+            'principal_investigator',
             'registered_identifier',
-            'title'
+            'title',
+            'data_modality'
         },
         'donor': {
             'donor_id',
             'organism_type',
             'phenotypic_sex',
             'reported_ethnicity',
+            'genetic_ancestry',
         },
         'file': {
             'file_id',
             'data_modality',
             'file_format',
             'file_size',
+            'file_md5sum',
             'reference_assembly',
+            'file_name',
             'file_ref',
         },
         'activity': {
@@ -531,12 +538,14 @@ def convert_column(value):
             'alignmentactivity_id',
             'activity_type',
             'data_modality',
+            'reference_assembly',
             # Not in schema
             'date_created',
         },
         'assayactivity': {
             'assayactivity_id',
             'activity_type',
+            'assay_type',
             'data_modality',
             # Not in schema
             'assay_category',
@@ -545,6 +554,7 @@ def convert_column(value):
         'sequencingactivity': {
             'sequencingactivity_id',
             'activity_type',
+            'assay_type',
             'data_modality',
         }
     }