Fix adapter aliasing for single-cell RNA-seq (fixes #8)

lobster/core/data_manager_v2.py

@@ -9,6 +9,7 @@
 import json
 import logging
 import os
+import re
 import tempfile
 import threading
 import time
@@ -754,6 +755,32 @@ def register_adapter(
         self.adapters[name] = adapter
         logger.debug(f"Registered adapter: {name} ({adapter.__class__.__name__})")
 
+    def _normalize_adapter_name(self, adapter: str) -> str:
+        """Normalize adapter names for tolerant registry lookup."""
+        normalized = adapter.lower().strip()
+        normalized = normalized.replace(" ", "_").replace("-", "_")
+        normalized = re.sub(r"_+", "_", normalized)
+        return normalized.strip("_")
+
+    def _resolve_adapter_name(self, adapter: str) -> str:
+        """Resolve aliases and normalized names to a registered adapter key."""
+        if adapter in self.adapters:
+            return adapter
+
+        normalized_adapter = self._normalize_adapter_name(adapter)
+        alias_map = {
+            "single_cell_rna_seq": "transcriptomics_single_cell",
+            "singlecell_rnaseq": "transcriptomics_single_cell",
+            "scrna": "transcriptomics_single_cell",
+            "scrna_seq": "transcriptomics_single_cell",
+        }
+
+        resolved_adapter = alias_map.get(normalized_adapter, normalized_adapter)
+        if resolved_adapter in self.adapters:
+            return resolved_adapter
+
+        raise ValueError(f"Adapter '{adapter}' not registered")
+
     def load_modality(
         self,
         name: str,
@@ -778,10 +805,8 @@ def load_modality(
         Raises:
             ValueError: If adapter is not registered or validation fails
         """
-        if adapter not in self.adapters:
-            raise ValueError(f"Adapter '{adapter}' not registered")
-
-        adapter_instance = self.adapters[adapter]
+        resolved_adapter = self._resolve_adapter_name(adapter)
+        adapter_instance = self.adapters[resolved_adapter]
 
         # Load data using adapter
         if not _is_anndata_instance(source):
@@ -815,7 +840,7 @@ def load_modality(
                 entity_type="modality_data",
                 metadata={
                     "modality_name": name,
-                    "adapter": adapter,
+                    "adapter": resolved_adapter,
                     "shape": adata.shape,
                 },
             )
@@ -835,18 +860,18 @@ def load_modality(
                 parameters={
                     "name": name,
                     "source": source_path,
-                    "adapter": adapter,
+                    "adapter": resolved_adapter,
                     **kwargs,
                 },
-                description=f"Loaded modality '{name}' using {adapter} adapter",
+                description=f"Loaded modality '{name}' using {resolved_adapter} adapter",
                 ir=loading_ir,
             )
 
             # Add provenance to AnnData
             adata = self.provenance.add_to_anndata(adata)
 
         logger.info(
-            f"Loaded modality '{name}': {adata.shape} using adapter '{adapter}'"
+            f"Loaded modality '{name}': {adata.shape} using adapter '{resolved_adapter}'"
         )
         return adata
 

tests/unit/core/test_adapter_aliases_issue8.py

@@ -0,0 +1,40 @@
+import pytest
+
+from lobster.core.data_manager_v2 import DataManagerV2
+
+
+@pytest.fixture
+def lightweight_data_manager():
+    dm = DataManagerV2.__new__(DataManagerV2)
+    dm.adapters = {"transcriptomics_single_cell": object()}
+    return dm
+
+
+@pytest.mark.unit
+@pytest.mark.parametrize(
+    "adapter_input",
+    [
+        "single_cell_rna_seq",
+        "single-cell rna-seq",
+        "scrna",
+        "scrna_seq",
+        "scRNA-seq",
+    ],
+)
+def test_single_cell_aliases_resolve_to_canonical(
+    lightweight_data_manager, adapter_input
+):
+    resolved = lightweight_data_manager._resolve_adapter_name(adapter_input)
+    assert resolved == "transcriptomics_single_cell"
+
+
+@pytest.mark.unit
+def test_canonical_adapter_resolves_as_is(lightweight_data_manager):
+    resolved = lightweight_data_manager._resolve_adapter_name("transcriptomics_single_cell")
+    assert resolved == "transcriptomics_single_cell"
+
+
+@pytest.mark.unit
+def test_invalid_adapter_still_raises_value_error(lightweight_data_manager):
+    with pytest.raises(ValueError, match="Adapter 'not_a_real_adapter' not registered"):
+        lightweight_data_manager._resolve_adapter_name("not_a_real_adapter")