Merge pull request #371 from broadinstitute/development

Release 1.37.0

.github/workflows/minify_ontologies.yml

@@ -3,9 +3,9 @@ name: Minify ontologies
 on:
   pull_request:
     types: [opened]  # Only trigger on PR "opened" event
-  push: # Uncomment, update branches to develop / debug
-    branches:
-      jb-metadata-boolean
+#  push: # Uncomment, update branches to develop / debug
+#    branches:
+#      jb-metadata-boolean
 
 jobs:
   build:

ingest/anndata_.py

@@ -187,7 +187,7 @@ def generate_processed_matrix(adata):
         h5ad_frag.features.processed.tsv
         Gzip files for faster delocalization
         """
-        if adata.var.index.name == 'gene_ids':
+        if AnnDataIngestor.check_ensembl_index(adata):
             # CELLxGENE indexes by Ensembl gene ID, not gene name (i.e. symbol).
             # Gene name is encoded in feature_name, which is needed for gene search.
             feature_frame = adata.var.feature_name
@@ -216,6 +216,15 @@ def generate_processed_matrix(adata):
         )
         AnnDataIngestor.compress_file(mtx_filename)
 
+    @staticmethod
+    def check_ensembl_index(adata):
+        """Check if an AnnData file is indexed on Ensembl gene IDs (e.g. ENSG00000243485) instead of gene symbols"""
+        if adata.var.index.name == 'gene_ids':
+            return True
+        else:
+            prefixes = list(set(gene_id[:3] for gene_id in adata.var_names))
+            return len(prefixes) == 1 and prefixes[0] == 'ENS'
+
     @staticmethod
     def delocalize_extracted_files(
         file_path, study_file_id, accession, files_to_delocalize

tests/test_anndata.py

@@ -266,6 +266,27 @@ def test_gene_id_indexed_generate_processed_matrix(self):
             first_line, expected_first_line, 'Expected Ensembl ID and gene name'
         )
 
+    def test_check_if_indexed_by_gene_id(self):
+        # check var.index.name
+        feature_name = AnnDataIngestor(
+            "../tests/data/anndata/indexed_by_gene_id.h5ad", self.study_id, self.study_file_id
+        )
+        adata = feature_name.obtain_adata()
+        self.assertTrue(feature_name.check_ensembl_index(adata))
+
+        # check data inspection
+        data_inspect = AnnDataIngestor(
+            "../tests/data/anndata/cellxgene.human_liver_b_cells.h5ad", self.study_id, self.study_file_id
+        )
+        liver_adata = data_inspect.obtain_adata()
+        self.assertTrue(data_inspect.check_ensembl_index(liver_adata))
+
+        # negative test
+        gene_symbols = AnnDataIngestor(
+            "../tests/data/anndata/anndata_test.h5ad", self.study_id, self.study_file_id
+        )
+        normal_adata = gene_symbols.obtain_adata()
+        self.assertFalse(gene_symbols.check_ensembl_index(normal_adata))
 
     def test_get_files_to_delocalize(self):
         files = AnnDataIngestor.clusterings_to_delocalize(self.valid_kwargs)