feat: add schema_reference field to uns during add-labels

cellxgene_schema_cli/cellxgene_schema/env.py

@@ -7,3 +7,5 @@
 PARSED_ONTOLOGIES_FILE = os.path.join(ONTOLOGY_DIR, "all_ontology.json.gz")
 SCHEMA_DEFINITIONS_DIR = os.path.join(PACKAGE_ROOT, "schema_definitions")
 SCHEMA_DEFINITION_FILE = os.path.join(SCHEMA_DEFINITIONS_DIR, "schema_definition.yaml")
+SCHEMA_REFERENCE_BASE_URL = "https://github.com/chanzuckerberg/single-cell-curation/blob/main/schema"
+SCHEMA_REFERENCE_FILE_NAME = "schema.md"

cellxgene_schema_cli/cellxgene_schema/schema_definitions/schema_definition.yaml

@@ -23,6 +23,7 @@ components:
         required: null # Means it's required
         reserved_columns:
             - schema_version
+            - schema_reference
         deprecated_columns:
             - X_normalization
             - default_field

cellxgene_schema_cli/cellxgene_schema/write_labels.py

@@ -5,6 +5,7 @@
 import pandas as pd
 
 from cellxgene_schema import ontology
+from cellxgene_schema.env import SCHEMA_REFERENCE_BASE_URL, SCHEMA_REFERENCE_FILE_NAME
 from cellxgene_schema.validate import ONTOLOGY_CHECKER, Validator
 
 from .utils import enforce_canonical_format, getattr_anndata
@@ -323,6 +324,9 @@ def _remove_categories_with_zero_values(self):
             if col.dtype == "category":
                 df[column] = col.cat.remove_unused_categories()
 
+    def _build_schema_reference_url(self, schema_version: str):
+        return f"{SCHEMA_REFERENCE_BASE_URL}/{schema_version}/{SCHEMA_REFERENCE_FILE_NAME}"
+
     def write_labels(self, add_labels_file: str):
         """
         From a valid (per cellxgene's schema) h5ad, this function writes a new h5ad file with ontology/gene labels added
@@ -341,6 +345,8 @@ def write_labels(self, add_labels_file: str):
 
         # Set version
         self.adata.uns["schema_version"] = self.validator.schema_version
+        # Set schema reference URL
+        self.adata.uns["schema_reference"] = self._build_schema_reference_url(self.validator.schema_version)
 
         enforce_canonical_format(self.adata)
 

cellxgene_schema_cli/tests/fixtures/examples_validate.py

@@ -162,6 +162,7 @@
 
 good_uns_with_labels = {
     "schema_version": "4.0.0",
+    "schema_reference": "https://github.com/chanzuckerberg/single-cell-curation/blob/main/schema/4.0.0/schema.md",
     "title": "A title",
     "default_embedding": "X_umap",
     "X_approximate_distribution": "normal",

cellxgene_schema_cli/tests/test_schema_compliance.py

@@ -1023,6 +1023,10 @@ def test_reserved_columns_presence(self):
 
         for reserved_column in self.validator.schema_def["components"]["uns"]["reserved_columns"]:
             with self.subTest(column=reserved_column):
+                # Resetting validator
+                self.validator.adata = examples.adata.copy()
+                self.validator.errors = []
+
                 self.validator.adata.uns[reserved_column] = "dummy_value"
                 self.validator.validate_adata()
                 self.assertEqual(