Add Stellaromics Pyxa (v1) xsmall dataset

pyxa_v1_io/README.md

@@ -0,0 +1,9 @@
+# Stellaromics Pyxa (v1 output files)
+
+The `xsmall` subset of the [Stellaromics/demo](https://huggingface.co/datasets/Stellaromics/demo) dataset on the
+Hugging Face Hub (BSD 3-Clause): a 100 × 100 × 100 µm cube (187 cells, ~23k transcripts) with the four Pyxa output
+files and a DAPI mosaic (multiscale OME-Zarr), ~8 MB in total. It is the same subset used in the spatialdata-io CI
+tests for the experimental `pyxa` reader.
+
+`download.py` fetches it into `data/`, and `to_zarr.py` converts it with `spatialdata_io.experimental.pyxa()`. The
+full demo region (`small/`, ~290 MB) has the same layout.

pyxa_v1_io/download.py

@@ -0,0 +1,32 @@
+##
+import os
+import subprocess
+import zipfile
+from pathlib import Path
+
+# from https://huggingface.co/datasets/Stellaromics/demo (BSD 3-Clause)
+# xsmall/: a 100 x 100 x 100 um cube cropped from the full demo region (small/), ~8 MB
+
+BASE_URL = "https://huggingface.co/datasets/Stellaromics/demo/resolve/main/xsmall"
+FILES = [
+    "cell_assigned_gene_v1.csv",
+    "cell_by_gene_v1.csv",
+    "cell_metadata_v1.csv",
+    "segmentation_geometries_v1.parquet",
+    "mosaic_3d.ome.zarr.zip",
+]
+
+data_dir = Path(__file__).resolve().parent / "data"
+os.makedirs(data_dir, exist_ok=True)
+
+##
+# download the data
+for filename in FILES:
+    command = f"curl -L -C - -o {data_dir / filename} {BASE_URL}/{filename}"
+    subprocess.run(command, shell=True, check=True)
+
+##
+# unzip the DAPI mosaic (a zipped OME-Zarr store containing mosaic_3d.ome.zarr/)
+with zipfile.ZipFile(data_dir / "mosaic_3d.ome.zarr.zip") as zf:
+    zf.extractall(data_dir)
+os.remove(data_dir / "mosaic_3d.ome.zarr.zip")

pyxa_v1_io/to_zarr.py

@@ -0,0 +1,40 @@
+# /// script
+# requires-python = ">=3.12"
+# dependencies = [
+#     # the pyxa reader is pending review in scverse/spatialdata-io
+#     "spatialdata-io @ git+https://github.com/ckmah/spatialdata-io.git@pyxa-reader",
+# ]
+# ///
+##
+from spatialdata_io.experimental import pyxa
+import spatialdata as sd
+
+##
+from pathlib import Path
+import shutil
+
+##
+path = Path().resolve()
+# luca's workaround for pycharm
+if not str(path).endswith("pyxa_v1_io"):
+    path /= "pyxa_v1_io"
+    assert path.exists()
+
+path_read = path / "data"
+path_write = path / "data.zarr"
+
+##
+print("parsing the data... ", end="")
+sdata = pyxa(path_read, image_path=path_read / "mosaic_3d.ome.zarr")
+print("done")
+
+##
+print("writing the data... ", end="")
+if path_write.exists():
+    shutil.rmtree(path_write)
+sdata.write(path_write)
+print("done")
+
+##
+sdata = sd.SpatialData.read(path_write)
+print(sdata)