Refactor utils into sdata + data subpackages and migrate sd_utils imports

pyproject.toml

@@ -49,8 +49,7 @@ dev = [
 where = ["src"]
 
 [tool.setuptools.package-data]
-xenium_analysis_tools = ["data/*.json"]
-"xenium_analysis_tools.utils" = ["*.csv", "*.json"]
+"xenium_analysis_tools.utils.data" = ["*.json", "reference_tables/*.csv"]
 
 [tool.setuptools.dynamic]
 version = {attr = "xenium_analysis_tools.__version__"}

src/xenium_analysis_tools/alignment/align_sections.py

@@ -35,16 +35,18 @@
     Image3DModel,
 )
 
-from xenium_analysis_tools.utils.sd_utils import (
+from xenium_analysis_tools.utils.sdata.transforms import (
     add_micron_coord_sys,
     get_microns_scales,
+    rename_coordinate_systems_manual,
+)
+from xenium_analysis_tools.utils.sdata.elements import (
     _is_multiscale,
     drop_sdata_elements,
     get_spatial_elements,
     get_single_scale,
-    separate_channels,
-    rename_coordinate_systems_manual
 )
+from xenium_analysis_tools.utils.sdata.channels import separate_channels
 
 def get_zstack_zarr(alignment_folder, paths, zstack_fov_size=None):
     zstack_folder = alignment_folder / 'zstacks'

src/xenium_analysis_tools/alignment/confocal.py

@@ -8,10 +8,8 @@
 import xarray as xr
 import spatialdata as sd
 from spatialdata.models import Image3DModel
-from xenium_analysis_tools.utils.sd_utils import (
-    add_micron_coord_sys,
-    write_sdata_elements
-)
+from xenium_analysis_tools.utils.sdata.transforms import add_micron_coord_sys
+from xenium_analysis_tools.utils.sdata.elements import write_sdata_elements
 
 def get_confocal_sdata(dataset_id,
                         confocal_path, 

src/xenium_analysis_tools/alignment/coregistration.py

@@ -18,10 +18,8 @@
 
 from spatialdata.transformations import get_transformation, Scale, Identity, Sequence
 
-from xenium_analysis_tools.utils.sd_utils import (
-    add_micron_coord_sys,
-    _is_multiscale,
-)
+from xenium_analysis_tools.utils.sdata.transforms import add_micron_coord_sys
+from xenium_analysis_tools.utils.sdata.elements import _is_multiscale
 
 from spatialdata.transformations import (
     get_transformation,

src/xenium_analysis_tools/alignment/qc_plots.py

@@ -35,9 +35,7 @@
     Image3DModel,
 )
 
-from xenium_analysis_tools.utils.sd_utils import (
-    add_micron_coord_sys,
-)
+from xenium_analysis_tools.utils.sdata.transforms import add_micron_coord_sys
 
 from xenium_analysis_tools.alignment.align_sections import (
     get_affine_from_landmarks_flat,

src/xenium_analysis_tools/alignment/xenium_alignment.py

@@ -5,10 +5,8 @@
 import matplotlib.pyplot as plt
 import xarray as xr
 
-from xenium_analysis_tools.utils.sd_utils import (
-    add_mapped_cells_cols,
-    add_micron_coord_sys
-)
+from xenium_analysis_tools.alignment.format_for_napari import add_mapped_cells_cols
+from xenium_analysis_tools.utils.sdata.transforms import add_micron_coord_sys
 
 from pathlib import Path
 import spatialdata as sd
@@ -312,7 +310,7 @@ def simplify_coord_systems(sdata, section_n):
     return sdata
 
 # import spatialdata as sd
-# from xenium_analysis_tools.utils.sd_utils import add_micron_coord_sys
+# from xenium_analysis_tools.utils.sdata.transforms import add_micron_coord_sys
 # from spatialdata.models import Image2DModel, Image3DModel, Labels3DModel, Labels2DModel, PointsModel, ShapesModel, TableModel
 # from spatialdata.transformations import get_transformation, set_transformation
 # import anndata as ad
@@ -324,10 +322,8 @@ def simplify_coord_systems(sdata, section_n):
 # import tifffile
 # import json
 # import re
-# from xenium_analysis_tools.utils.sd_utils import (
-#     add_mapped_cells_cols,
-#     get_transcripts_bboxes
-# )
+# from xenium_analysis_tools.alignment.format_for_napari import add_mapped_cells_cols
+# from xenium_analysis_tools.utils.sdata.elements import get_transcripts_bboxes
 
 # def create_zstack_array(tif_path,
 #                         add_chan=True,

src/xenium_analysis_tools/alignment/zstack_alignment.py

@@ -3,7 +3,7 @@
 import xarray as xr
 from spatialdata.models import Image3DModel, Labels3DModel, TableModel
 from spatialdata.transformations import Identity, Scale, set_transformation
-from xenium_analysis_tools.utils.sd_utils import add_micron_coord_sys
+from xenium_analysis_tools.utils.sdata.transforms import add_micron_coord_sys
 import anndata as ad
 from pathlib import Path
 import spatialdata as sd

src/xenium_analysis_tools/example_notebooks/combine_sections.ipynb

@@ -18,7 +18,7 @@
     "import gc\n",
     "import os\n",
     "\n",
-    "from xenium_analysis_tools.utils.sd_utils import (\n",
+    "from xenium_analysis_tools.utils.sdata.file_ops import (\n",
     "    get_dataset_paths,\n",
     ")\n",
     "\n",

src/xenium_analysis_tools/example_notebooks/map_dataset.ipynb

@@ -35,7 +35,7 @@
     "    load_config, \n",
     ")\n",
     "\n",
-    "from xenium_analysis_tools.utils.sd_utils import (\n",
+    "from xenium_analysis_tools.utils.sdata.file_ops import (\n",
     "    get_dataset_paths\n",
     ")\n",
     "\n",

src/xenium_analysis_tools/process_sections/process_xenium_bundles.py

@@ -46,13 +46,9 @@
     Sequence,
 )
 
-from xenium_analysis_tools.utils.sd_utils import (
-    extract_scale_transform,
-    write_sdata_elements,
-    add_micron_coord_sys,
-    get_ome_metadata,
-    rename_chans,
-)
+from xenium_analysis_tools.utils.sdata.transforms import extract_scale_transform, add_micron_coord_sys
+from xenium_analysis_tools.utils.sdata.elements import write_sdata_elements, get_ome_metadata
+from xenium_analysis_tools.utils.sdata.channels import rename_chans
 
 
 ##############################

src/xenium_analysis_tools/utils/__init__.py

@@ -1,25 +1,59 @@
 from importlib.resources import files as _files
 from pathlib import Path
 
+from .env import detect_env, get_datasets_json_path
+from .sdata import (
+    add_micron_coord_sys,
+    cleanup_core_dumps,
+    drop_sdata_elements,
+    extract_scale_transform,
+    get_channel_name,
+    get_dataset_paths,
+    get_element_bytes,
+    get_microns_scales,
+    get_ome_metadata,
+    get_single_scale,
+    get_spatial_elements,
+    get_transcripts_bboxes,
+    load_dataset_json,
+    move_bundle_sections,
+    move_sections_to_final_folder,
+    print_sdata_size_summary,
+    rename_chans,
+    rename_coordinate_systems_manual,
+    rename_elements_section,
+    separate_channels,
+    write_sdata_elements,
+)
 
-def get_datasets_info_file(filename: str) -> Path:
-    """Return a Path to a data file bundled in xenium_analysis_tools/utils/.
-
-    Parameters
-    ----------
-    filename : str
-        Name of the file, e.g. "xenium_datasets.json" or "my_data.csv".
+__all__ = [
+    "add_micron_coord_sys",
+    "cleanup_core_dumps",
+    "detect_env",
+    "drop_sdata_elements",
+    "extract_scale_transform",
+    "get_channel_name",
+    "get_dataset_paths",
+    "get_datasets_info_file",
+    "get_datasets_json_path",
+    "get_element_bytes",
+    "get_microns_scales",
+    "get_ome_metadata",
+    "get_single_scale",
+    "get_spatial_elements",
+    "get_transcripts_bboxes",
+    "load_dataset_json",
+    "move_bundle_sections",
+    "move_sections_to_final_folder",
+    "print_sdata_size_summary",
+    "rename_chans",
+    "rename_coordinate_systems_manual",
+    "rename_elements_section",
+    "separate_channels",
+    "write_sdata_elements",
+]
 
-    Returns
-    -------
-    Path
-        Resolved path to the file inside the installed package.
 
-    Example
-    -------
-    >>> import json
-    >>> from xenium_analysis_tools.utils import get_utils_file
-    >>> with open(get_utils_file("xenium_datasets.json")) as f:
-    ...     data = json.load(f)
-    """
-    return Path(str(_files("xenium_analysis_tools.utils").joinpath(filename)))
+def get_datasets_info_file(filename: str) -> Path:
+    """Return a Path to a data file bundled in xenium_analysis_tools/utils/data/."""
+    return Path(str(_files("xenium_analysis_tools.utils.data").joinpath(filename)))

src/xenium_analysis_tools/utils/env.py

@@ -35,4 +35,4 @@ def detect_env(root_path=None, data_root=None, scratch_root=None, results_root=N
 
 def get_datasets_json_path():
     """Return a Path to the bundled xenium_datasets.json inside the package."""
-    return _files("xenium_analysis_tools.data").joinpath("xenium_datasets.json")
+    return _files("xenium_analysis_tools.utils.data").joinpath("xenium_datasets.json")