Fix a bug in scdl header saving. Files were saved to data_path, but this was occasionally changed over the lifetime of an object, leading to issues where the header_path was not updated to track the new datapath.

Signed-off-by: Eric T. Dawson <edawson@nvidia.com>

Author: Eric T. Dawson

sub-packages/bionemo-scdl/pyproject.toml

@@ -22,7 +22,7 @@ dependencies = [
 ]
 
 [project.optional-dependencies]
-dev = [
+test = [
     "bionemo-core>=2.4.0",
     'pytest>=8.4.1'
 ]

sub-packages/bionemo-scdl/src/bionemo/scdl/io/single_cell_memmap_dataset.py

@@ -248,7 +248,7 @@ def __init__(
         """
         self._version: str = importlib.metadata.version("bionemo.scdl")
         self.data_path: str = data_path
-        self.header_path: Path = Path(data_path) / "header.sch"
+        self.header_file_name: str = "header.sch"
         self.header: SCDLHeader = None
         self.mode: Mode = mode
         self.paginated_load_cutoff = paginated_load_cutoff
@@ -708,13 +708,13 @@ def load(self, stored_path: str) -> None:
             )
         self.data_path = stored_path
         self.mode = Mode.READ_APPEND
-        self.header_path = Path(stored_path) / "header.sch"
+        # self.header_path = Path(stored_path) / self.header_file_name
         # Load header if present; keep None if missing or unreadable
-        if os.path.exists(self.header_path):
+        if os.path.exists(self.data_path / self.header_file_name):
             try:
-                self.header = SCDLHeader.load(str(self.header_path))
+                self.header = SCDLHeader.load(str(self.data_path / self.header_file_name))
             except Exception as e:
-                warnings.warn(f"Failed to load SCDL header at {self.header_path}: {e}")
+                warnings.warn(f"Failed to load SCDL header at {Path(self.data_path) / self.header_file_name}: {e}")
                 self.header = None
         else:
             self.header = None
@@ -1024,7 +1024,7 @@ def _write_header(self):
                 indexes,
             )
         )
-        header.save(str(self.header_path))
+        header.save(Path(self.data_path) / self.header_file_name)
 
     def save(self, output_path: Optional[str] = None) -> None:
         """Saves the class to a given output path.

sub-packages/bionemo-scdl/tests/bionemo/scdl/conftest.py

@@ -29,7 +29,6 @@ def test_directory() -> Path:
     Returns:
         A Path object that is the directory with test data.
     """
-    # return load("scdl/sample") / "scdl_data"
     return load("scdl/sample_scdl_feature_ids") / "scdl_data_with_feature_ids"