patchworks 0.2.0__tar.gz → 0.3.0__tar.gz

patchworks-0.3.0/.github/workflows/lint.yml

@@ -0,0 +1,23 @@
+name: Lint
+
+on:
+  push:
+    branches:
+      - main
+  pull_request:
+
+jobs:
+  ruff:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: ruff check
+        uses: astral-sh/ruff-action@v3
+        with:
+          args: "check"
+
+      - name: ruff format --check
+        uses: astral-sh/ruff-action@v3
+        with:
+          args: "format --check"

{patchworks-0.2.0 → patchworks-0.3.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: patchworks
-Version: 0.2.0
+Version: 0.3.0
 Summary: Tiled processing of arbitrarily large images with globally consistent labels
 Project-URL: Homepage, https://github.com/imcf/patchworks
 Project-URL: Issues, https://github.com/imcf/patchworks/issues
@@ -22,10 +22,15 @@ Requires-Dist: numpy>=1.24
 Requires-Dist: scipy>=1.9
 Requires-Dist: zarr>=2.14
 Provides-Extra: all
+Requires-Dist: bioio; extra == 'all'
+Requires-Dist: bioio-bioformats; extra == 'all'
 Requires-Dist: nvidia-ml-py; extra == 'all'
 Requires-Dist: psutil; extra == 'all'
 Requires-Dist: scikit-image; extra == 'all'
 Requires-Dist: tqdm; extra == 'all'
+Provides-Extra: bioio
+Requires-Dist: bioio; extra == 'bioio'
+Requires-Dist: bioio-bioformats; extra == 'bioio'
 Provides-Extra: cellpose
 Requires-Dist: cellpose>=3.0; extra == 'cellpose'
 Provides-Extra: dev
@@ -42,10 +47,17 @@ Requires-Dist: nvidia-ml-py; extra == 'gpu'
 Provides-Extra: io
 Requires-Dist: psutil; extra == 'io'
 Requires-Dist: tqdm; extra == 'io'
+Provides-Extra: napari
+Requires-Dist: napari[all]; extra == 'napari'
 Description-Content-Type: text/markdown
 
 # patchworks
 
+[![PyPI](https://img.shields.io/pypi/v/patchworks.svg)](https://pypi.org/project/patchworks/)
+[![Python versions](https://img.shields.io/pypi/pyversions/patchworks.svg)](https://pypi.org/project/patchworks/)
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
+[![Docs](https://img.shields.io/badge/docs-imcf.one%2Fpatchworks-blue)](https://imcf.one/patchworks/)
+
 > Tiled processing of arbitrarily large images — any image, any function.
 
 ```
@@ -75,9 +87,15 @@ Optional extras:
 ```bash
 pip install "patchworks[gpu]"      # GPU VRAM querying (nvidia-ml-py)
 pip install "patchworks[cellpose]" # Cellpose plugin
-pip install "patchworks[all]"      # Everything
+pip install "patchworks[bioio]"    # convert any image format to OME-ZARR
+pip install "patchworks[napari]"   # interactive napari viewer plugin
+pip install "patchworks[all]"      # Everything (except napari GUI)
 ```
 
+> `bioio` reads CZI/LIF/ND2/OME-TIFF/… — it ships with `bioio-bioformats`, the
+> Bio-Formats catch-all reader (needs a JVM). Add faster native readers where
+> you want them (e.g. `bioio-ome-tiff`, `bioio-czi`, `bioio-lif`, `bioio-nd2`).
+
 ---
 
 ## Quick start — 5 lines
@@ -85,11 +103,14 @@ pip install "patchworks[all]"      # Everything
 ```python
 from patchworks import tile_process
 
+
 def my_fn(tile):
     from skimage.filters import threshold_otsu
     from skimage.measure import label
+
     return label(tile > threshold_otsu(tile)).astype("int32")
 
+
 result = tile_process("image.zarr", my_fn, compute=True)
 ```
 
@@ -107,10 +128,11 @@ from patchworks.plugins.cellpose import cellpose_fn
 fn = cellpose_fn("cyto3", gpu=True, diameter=30)
 
 tile_process(
-    "image.zarr", fn,
+    "image.zarr",
+    fn,
     tile_shape=(1, 2048, 2048),  # one z-slice per tile
-    overlap=20,                  # gives boundary cells enough context
-    write_to="labels.zarr",      # stream directly to disk — no RAM accumulation
+    overlap=20,  # gives boundary cells enough context
+    write_to="labels.zarr",  # stream directly to disk — no RAM accumulation
     progress=True,
 )
 ```
@@ -125,15 +147,22 @@ from patchworks import tile_process
 
 model = StarDist2D.from_pretrained("2D_versatile_fluo")
 
+
 def stardist_fn(tile):
     img = tile[0] if tile.ndim == 3 and tile.shape[0] == 1 else tile
     norm = img.astype("float32") / (img.max() or 1)
     labels, _ = model.predict_instances(norm)
     return labels.astype("int32")[None] if tile.ndim == 3 else labels.astype("int32")
 
-tile_process("image.zarr", stardist_fn,
-             tile_shape=(1, 1024, 1024), overlap=32,
-             write_to="labels.zarr", progress=True)
+
+tile_process(
+    "image.zarr",
+    stardist_fn,
+    tile_shape=(1, 1024, 1024),
+    overlap=32,
+    write_to="labels.zarr",
+    progress=True,
+)
 ```
 
 ---
@@ -146,11 +175,13 @@ from scipy.ndimage import gaussian_filter
 from skimage.measure import label
 from patchworks import tile_process
 
+
 def my_custom_fn(tile: np.ndarray) -> np.ndarray:
     smoothed = gaussian_filter(tile.astype("float32"), sigma=1.5)
     binary = smoothed > smoothed.mean()
     return label(binary).astype("int32")
 
+
 tile_process("image.zarr", my_custom_fn, tile_shape=(1, 512, 512))
 ```
 
@@ -174,11 +205,14 @@ from patchworks import estimate_empty_tiles, tile_process
 info = estimate_empty_tiles("image.zarr", tile_shape=(120, 697, 697))
 print(f"{info['empty_fraction']:.0%} tiles are background — will be skipped")
 
-tile_process("image.zarr", fn,
-             tile_shape=(120, 697, 697),
-             skip_empty=True,
-             empty_threshold=info["threshold"],
-             write_to="labels.zarr")
+tile_process(
+    "image.zarr",
+    fn,
+    tile_shape=(120, 697, 697),
+    skip_empty=True,
+    empty_threshold=info["threshold"],
+    write_to="labels.zarr",
+)
 ```
 
 ### Distributed cluster for GPU
@@ -190,7 +224,8 @@ client, cluster = make_local_cluster(use_gpu=True)
 try:
     tile_process("image.zarr", fn, write_to="labels.zarr", progress=True)
 finally:
-    client.close(); cluster.close()
+    client.close()
+    cluster.close()
 ```
 
 ### Contiguous label numbering
@@ -198,9 +233,7 @@ finally:
 ```python
 # Labels are globally unique by default, but may be gappy (block-encoded IDs).
 # sequential_labels=True does a linear relabel O(voxels) — not O(n_tiles²).
-tile_process("image.zarr", fn,
-             write_to="labels.zarr",
-             sequential_labels=True)
+tile_process("image.zarr", fn, write_to="labels.zarr", sequential_labels=True)
 ```
 
 ### Use only the merge step (bring your own tiling)
@@ -215,8 +248,9 @@ from patchworks import merge_tile_labels
 
 # Your own tiling + segmentation
 image = da.from_zarr("image.zarr").rechunk((1, 1024, 1024))
-labeled = image.map_blocks(my_segment_fn, dtype="int32",
-                            meta=np.empty((0,) * image.ndim, dtype="int32"))
+labeled = image.map_blocks(
+    my_segment_fn, dtype="int32", meta=np.empty((0,) * image.ndim, dtype="int32")
+)
 
 merged = merge_tile_labels(labeled, write_to="labels.zarr", progress=True)
 ```
@@ -257,13 +291,13 @@ tiles where the dask-image approach stalls.
 
 ## Known pitfalls (and how patchworks avoids them)
 
-| Pitfall | Symptom | How patchworks handles it |
-|---|---|---|
-| In-process Dask client | `FutureCancelledError: lost dependencies` | Detected at startup, raises immediately with fix instructions |
-| 3-4× fn recompute during merge | Cellpose runs 3× per tile | Staging writes labels once, merge reads from disk |
-| O(n²) sequential relabelling | Graph construction hangs at 1000+ tiles | Linear post-pass O(voxels) via `np.unique` + LUT |
-| Wrong overlap boundary | Output shape mismatch | Always uses `boundary="none"` |
-| Persisting large arrays | Worker OOM | Never persists; keeps dask graph lazy and streams |
+| Pitfall                        | Symptom                                   | How patchworks handles it                                     |
+| ------------------------------ | ----------------------------------------- | ------------------------------------------------------------- |
+| In-process Dask client         | `FutureCancelledError: lost dependencies` | Detected at startup, raises immediately with fix instructions |
+| 3-4× fn recompute during merge | Cellpose runs 3× per tile                 | Staging writes labels once, merge reads from disk             |
+| O(n²) sequential relabelling   | Graph construction hangs at 1000+ tiles   | Linear post-pass O(voxels) via `np.unique` + LUT              |
+| Wrong overlap boundary         | Output shape mismatch                     | Always uses `boundary="none"`                                 |
+| Persisting large arrays        | Worker OOM                                | Never persists; keeps dask graph lazy and streams             |
 
 ---
 

{patchworks-0.2.0 → patchworks-0.3.0}/README.md

@@ -1,5 +1,10 @@
 # patchworks
 
+[![PyPI](https://img.shields.io/pypi/v/patchworks.svg)](https://pypi.org/project/patchworks/)
+[![Python versions](https://img.shields.io/pypi/pyversions/patchworks.svg)](https://pypi.org/project/patchworks/)
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
+[![Docs](https://img.shields.io/badge/docs-imcf.one%2Fpatchworks-blue)](https://imcf.one/patchworks/)
+
 > Tiled processing of arbitrarily large images — any image, any function.
 
 ```
@@ -29,9 +34,15 @@ Optional extras:
 ```bash
 pip install "patchworks[gpu]"      # GPU VRAM querying (nvidia-ml-py)
 pip install "patchworks[cellpose]" # Cellpose plugin
-pip install "patchworks[all]"      # Everything
+pip install "patchworks[bioio]"    # convert any image format to OME-ZARR
+pip install "patchworks[napari]"   # interactive napari viewer plugin
+pip install "patchworks[all]"      # Everything (except napari GUI)
 ```
 
+> `bioio` reads CZI/LIF/ND2/OME-TIFF/… — it ships with `bioio-bioformats`, the
+> Bio-Formats catch-all reader (needs a JVM). Add faster native readers where
+> you want them (e.g. `bioio-ome-tiff`, `bioio-czi`, `bioio-lif`, `bioio-nd2`).
+
 ---
 
 ## Quick start — 5 lines
@@ -39,11 +50,14 @@ pip install "patchworks[all]"      # Everything
 ```python
 from patchworks import tile_process
 
+
 def my_fn(tile):
     from skimage.filters import threshold_otsu
     from skimage.measure import label
+
     return label(tile > threshold_otsu(tile)).astype("int32")
 
+
 result = tile_process("image.zarr", my_fn, compute=True)
 ```
 
@@ -61,10 +75,11 @@ from patchworks.plugins.cellpose import cellpose_fn
 fn = cellpose_fn("cyto3", gpu=True, diameter=30)
 
 tile_process(
-    "image.zarr", fn,
+    "image.zarr",
+    fn,
     tile_shape=(1, 2048, 2048),  # one z-slice per tile
-    overlap=20,                  # gives boundary cells enough context
-    write_to="labels.zarr",      # stream directly to disk — no RAM accumulation
+    overlap=20,  # gives boundary cells enough context
+    write_to="labels.zarr",  # stream directly to disk — no RAM accumulation
     progress=True,
 )
 ```
@@ -79,15 +94,22 @@ from patchworks import tile_process
 
 model = StarDist2D.from_pretrained("2D_versatile_fluo")
 
+
 def stardist_fn(tile):
     img = tile[0] if tile.ndim == 3 and tile.shape[0] == 1 else tile
     norm = img.astype("float32") / (img.max() or 1)
     labels, _ = model.predict_instances(norm)
     return labels.astype("int32")[None] if tile.ndim == 3 else labels.astype("int32")
 
-tile_process("image.zarr", stardist_fn,
-             tile_shape=(1, 1024, 1024), overlap=32,
-             write_to="labels.zarr", progress=True)
+
+tile_process(
+    "image.zarr",
+    stardist_fn,
+    tile_shape=(1, 1024, 1024),
+    overlap=32,
+    write_to="labels.zarr",
+    progress=True,
+)
 ```
 
 ---
@@ -100,11 +122,13 @@ from scipy.ndimage import gaussian_filter
 from skimage.measure import label
 from patchworks import tile_process
 
+
 def my_custom_fn(tile: np.ndarray) -> np.ndarray:
     smoothed = gaussian_filter(tile.astype("float32"), sigma=1.5)
     binary = smoothed > smoothed.mean()
     return label(binary).astype("int32")
 
+
 tile_process("image.zarr", my_custom_fn, tile_shape=(1, 512, 512))
 ```
 
@@ -128,11 +152,14 @@ from patchworks import estimate_empty_tiles, tile_process
 info = estimate_empty_tiles("image.zarr", tile_shape=(120, 697, 697))
 print(f"{info['empty_fraction']:.0%} tiles are background — will be skipped")
 
-tile_process("image.zarr", fn,
-             tile_shape=(120, 697, 697),
-             skip_empty=True,
-             empty_threshold=info["threshold"],
-             write_to="labels.zarr")
+tile_process(
+    "image.zarr",
+    fn,
+    tile_shape=(120, 697, 697),
+    skip_empty=True,
+    empty_threshold=info["threshold"],
+    write_to="labels.zarr",
+)
 ```
 
 ### Distributed cluster for GPU
@@ -144,7 +171,8 @@ client, cluster = make_local_cluster(use_gpu=True)
 try:
     tile_process("image.zarr", fn, write_to="labels.zarr", progress=True)
 finally:
-    client.close(); cluster.close()
+    client.close()
+    cluster.close()
 ```
 
 ### Contiguous label numbering
@@ -152,9 +180,7 @@ finally:
 ```python
 # Labels are globally unique by default, but may be gappy (block-encoded IDs).
 # sequential_labels=True does a linear relabel O(voxels) — not O(n_tiles²).
-tile_process("image.zarr", fn,
-             write_to="labels.zarr",
-             sequential_labels=True)
+tile_process("image.zarr", fn, write_to="labels.zarr", sequential_labels=True)
 ```
 
 ### Use only the merge step (bring your own tiling)
@@ -169,8 +195,9 @@ from patchworks import merge_tile_labels
 
 # Your own tiling + segmentation
 image = da.from_zarr("image.zarr").rechunk((1, 1024, 1024))
-labeled = image.map_blocks(my_segment_fn, dtype="int32",
-                            meta=np.empty((0,) * image.ndim, dtype="int32"))
+labeled = image.map_blocks(
+    my_segment_fn, dtype="int32", meta=np.empty((0,) * image.ndim, dtype="int32")
+)
 
 merged = merge_tile_labels(labeled, write_to="labels.zarr", progress=True)
 ```
@@ -211,13 +238,13 @@ tiles where the dask-image approach stalls.
 
 ## Known pitfalls (and how patchworks avoids them)
 
-| Pitfall | Symptom | How patchworks handles it |
-|---|---|---|
-| In-process Dask client | `FutureCancelledError: lost dependencies` | Detected at startup, raises immediately with fix instructions |
-| 3-4× fn recompute during merge | Cellpose runs 3× per tile | Staging writes labels once, merge reads from disk |
-| O(n²) sequential relabelling | Graph construction hangs at 1000+ tiles | Linear post-pass O(voxels) via `np.unique` + LUT |
-| Wrong overlap boundary | Output shape mismatch | Always uses `boundary="none"` |
-| Persisting large arrays | Worker OOM | Never persists; keeps dask graph lazy and streams |
+| Pitfall                        | Symptom                                   | How patchworks handles it                                     |
+| ------------------------------ | ----------------------------------------- | ------------------------------------------------------------- |
+| In-process Dask client         | `FutureCancelledError: lost dependencies` | Detected at startup, raises immediately with fix instructions |
+| 3-4× fn recompute during merge | Cellpose runs 3× per tile                 | Staging writes labels once, merge reads from disk             |
+| O(n²) sequential relabelling   | Graph construction hangs at 1000+ tiles   | Linear post-pass O(voxels) via `np.unique` + LUT              |
+| Wrong overlap boundary         | Output shape mismatch                     | Always uses `boundary="none"`                                 |
+| Persisting large arrays        | Worker OOM                                | Never persists; keeps dask graph lazy and streams             |
 
 ---
 

patchworks-0.3.0/docs/api/plugins/napari.md

@@ -0,0 +1,7 @@
+# napari viewer plugin
+
+Open an OME-ZARR image and overlay the labels produced by
+[`tile_process`](../tile_process.md) as a napari *Labels* layer in a single
+call. Requires the optional `napari` extra (`pip install "patchworks[napari]"`).
+
+::: patchworks.plugins.napari.view_in_napari

patchworks-0.3.0/docs/api/plugins/ome_zarr.md

@@ -0,0 +1,26 @@
+# OME-ZARR conversion plugin
+
+Write any array or image file to a pyramidal OME-ZARR store, add resolution
+levels to an existing store, or store a label image inside an OME-ZARR under
+the NGFF `labels/` group. Uses only the core dependencies for arrays and
+`.zarr` inputs; reading other file formats needs the optional `bioio` extra
+(`pip install "patchworks[bioio]"`).
+
+Pyramids downsample **X and Y only** — `Z` (and channel/time) are kept at full
+resolution, matching anisotropic microscopy stacks.
+
+## to_ome_zarr
+
+::: patchworks.plugins.ome_zarr.to_ome_zarr
+
+## add_pyramid
+
+::: patchworks.plugins.ome_zarr.add_pyramid
+
+## write_labels
+
+::: patchworks.plugins.ome_zarr.write_labels
+
+## register_labels
+
+::: patchworks.plugins.ome_zarr.register_labels

{patchworks-0.2.0 → patchworks-0.3.0}/docs/examples/cellpose_2d.md

@@ -18,8 +18,8 @@ from patchworks.plugins.cellpose import cellpose_fn
 
 IMAGE = "image.zarr"
 OUTPUT = "labels.zarr"
-CHANNEL = 0       # channel to segment
-DIAMETER = 30     # expected cell diameter in pixels
+CHANNEL = 0  # channel to segment
+DIAMETER = 30  # expected cell diameter in pixels
 
 # 1. Create the Cellpose function
 fn = cellpose_fn("cyto3", gpu=True, diameter=DIAMETER)
@@ -31,10 +31,11 @@ print(f"{info['empty_fraction']:.0%} of slices are background")
 # 3. Run
 tile_fn = partial(auto_tile_shape_cellpose, diameter=DIAMETER, use_gpu=True)
 tile_process(
-    IMAGE, fn,
+    IMAGE,
+    fn,
     channel=CHANNEL,
-    tile_shape=tile_fn,       # auto-sized for GPU VRAM
-    overlap=20,               # 20-voxel halo for boundary cells
+    tile_shape=tile_fn,  # auto-sized for GPU VRAM
+    overlap=20,  # 20-voxel halo for boundary cells
     skip_empty=True,
     empty_threshold=info["threshold"],
     write_to=OUTPUT,

{patchworks-0.2.0 → patchworks-0.3.0}/docs/examples/cellpose_2d.py

@@ -3,6 +3,7 @@
 Each z-slice is processed independently. Tiles overlap by 20 voxels so cells
 near tile boundaries are fully visible to Cellpose.
 """
+
 from functools import partial
 
 from patchworks import auto_tile_shape_cellpose, tile_process
@@ -19,11 +20,12 @@ fn = cellpose_fn("cyto3", gpu=True, diameter=DIAMETER)
 tile_fn = partial(auto_tile_shape_cellpose, diameter=DIAMETER, use_gpu=True)
 
 tile_process(
-    IMAGE, fn,
+    IMAGE,
+    fn,
     channel=CHANNEL,
     tile_shape=tile_fn,
     overlap=20,
-    skip_empty=True,        # skip background slices
+    skip_empty=True,  # skip background slices
     write_to=OUTPUT,
     progress=True,
 )

{patchworks-0.2.0 → patchworks-0.3.0}/docs/examples/cellpose_3d.md

@@ -18,7 +18,7 @@ from patchworks.plugins.cellpose import cellpose_fn
 IMAGE = "image.zarr"
 OUTPUT = "labels_3d.zarr"
 CHANNEL = 0
-DIAMETER = 20     # pixels
+DIAMETER = 20  # pixels
 ANISOTROPY = 3.0  # z_spacing / xy_spacing
 
 fn = cellpose_fn(
@@ -45,7 +45,8 @@ print("Dashboard:", client.dashboard_link)
 
 try:
     tile_process(
-        IMAGE, fn,
+        IMAGE,
+        fn,
         channel=CHANNEL,
         tile_shape=tile_fn,
         overlap=10,

{patchworks-0.2.0 → patchworks-0.3.0}/docs/examples/cellpose_3d.py

@@ -3,15 +3,20 @@
 Each tile contains the full z extent. Cellpose do_3D=True runs segmentation on
 xy, xz, and yz planes and takes a 3-D consensus.
 """
+
 from functools import partial
 
-from patchworks import auto_tile_shape_cellpose, make_local_cluster, tile_process
+from patchworks import (
+    auto_tile_shape_cellpose,
+    make_local_cluster,
+    tile_process,
+)
 from patchworks.plugins.cellpose import cellpose_fn
 
 IMAGE = "image.zarr"
 OUTPUT = "labels_3d.zarr"
 CHANNEL = 0
-DIAMETER = 20   # pixels
+DIAMETER = 20  # pixels
 ANISOTROPY = 3.0  # z-spacing / xy-spacing
 
 fn = cellpose_fn(
@@ -35,7 +40,8 @@ print("Dashboard:", client.dashboard_link)
 
 try:
     tile_process(
-        IMAGE, fn,
+        IMAGE,
+        fn,
         channel=CHANNEL,
         tile_shape=tile_fn,
         overlap=10,

{patchworks-0.2.0 → patchworks-0.3.0}/docs/examples/custom.md

@@ -11,10 +11,12 @@ from skimage.filters import threshold_otsu
 from skimage.measure import label
 from patchworks import tile_process
 
+
 def threshold_fn(tile: np.ndarray) -> np.ndarray:
     thr = threshold_otsu(tile)
     return label(tile > thr).astype("int32")
 
+
 result = tile_process("image.zarr", threshold_fn, compute=True)
 ```
 
@@ -27,17 +29,22 @@ from skimage.morphology import remove_small_objects
 from skimage.measure import label
 from patchworks import tile_process
 
+
 def smooth_and_label(tile: np.ndarray) -> np.ndarray:
     smoothed = gaussian_filter(tile.astype("float32"), sigma=1.5)
     binary = smoothed > smoothed.mean()
     cleaned = remove_small_objects(binary, min_size=100)
     return label(cleaned).astype("int32")
 
-tile_process("image.zarr", smooth_and_label,
-             tile_shape=(1, 512, 512),
-             overlap=16,
-             write_to="labels.zarr",
-             progress=True)
+
+tile_process(
+    "image.zarr",
+    smooth_and_label,
+    tile_shape=(1, 512, 512),
+    overlap=16,
+    write_to="labels.zarr",
+    progress=True,
+)
 ```
 
 ## PyTorch model
@@ -51,6 +58,7 @@ from patchworks import tile_process
 device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
 model = MySegmentationModel().to(device).eval()
 
+
 @torch.no_grad()
 def torch_fn(tile: np.ndarray) -> np.ndarray:
     t = torch.from_numpy(tile.astype("float32")).unsqueeze(0).unsqueeze(0).to(device)
@@ -58,11 +66,15 @@ def torch_fn(tile: np.ndarray) -> np.ndarray:
     pred = logits.argmax(1).squeeze(0).cpu().numpy()
     return pred.astype("int32")
 
-tile_process("image.zarr", torch_fn,
-             tile_shape=(1, 512, 512),
-             use_gpu=True,
-             write_to="labels.zarr",
-             progress=True)
+
+tile_process(
+    "image.zarr",
+    torch_fn,
+    tile_shape=(1, 512, 512),
+    use_gpu=True,
+    write_to="labels.zarr",
+    progress=True,
+)
 ```
 
 ## Any array input, not just zarr
@@ -79,6 +91,7 @@ result = tile_process(arr, my_fn, compute=True)
 # From tifffile
 import tifffile
 import dask.array as da
+
 arr = da.from_array(tifffile.imread("image.tif", aszarr=True))
 result = tile_process(arr, my_fn, compute=True)
 ```

{patchworks-0.2.0 → patchworks-0.3.0}/docs/examples/custom_method.py

@@ -3,6 +3,7 @@
 patchworks doesn't care what's inside fn. Here's a more elaborate example
 using scipy + skimage with preprocessing.
 """
+
 import numpy as np
 from scipy.ndimage import gaussian_filter
 from skimage.filters import threshold_otsu

{patchworks-0.2.0 → patchworks-0.3.0}/docs/examples/standalone_merge.md

@@ -13,12 +13,15 @@ from patchworks import merge_tile_labels
 # Your own tiling + segmentation
 image = da.from_zarr("image.zarr").rechunk((1, 1024, 1024))
 
+
 def my_fn(tile: np.ndarray) -> np.ndarray:
     ...  # your segmentation code
     return labels.astype("int32")
 
+
 labeled = image.map_blocks(
-    my_fn, dtype="int32",
+    my_fn,
+    dtype="int32",
     meta=np.empty((0,) * image.ndim, dtype="int32"),
 )
 
@@ -37,7 +40,7 @@ merged = merge_tile_labels(
     "my_staged_labels.zarr",
     input_component="raw_labels",  # component name inside the zarr
     write_to="merged.zarr",
-    sequential_labels=True,        # renumber to 1..N
+    sequential_labels=True,  # renumber to 1..N
 )
 ```
 
@@ -64,5 +67,6 @@ use `merge_tile_labels`:
 labeled = your_pipeline(da.from_zarr("image.zarr"))  # dask.array.Array
 
 from patchworks import merge_tile_labels
+
 merged = merge_tile_labels(labeled, write_to="final.zarr")
 ```

{patchworks-0.2.0 → patchworks-0.3.0}/docs/examples/stardist.md

@@ -33,10 +33,11 @@ def stardist_fn(tile: np.ndarray) -> np.ndarray:
 
 
 tile_process(
-    IMAGE, stardist_fn,
+    IMAGE,
+    stardist_fn,
     channel=0,
     tile_shape=(1, 1024, 1024),
-    overlap=32,          # StarDist receptive field is larger than Cellpose
+    overlap=32,  # StarDist receptive field is larger than Cellpose
     write_to=OUTPUT,
     progress=True,
 )
@@ -51,10 +52,12 @@ tile_process(
     ```python
     from functools import lru_cache
 
+
     @lru_cache(maxsize=1)
     def _get_model():
         return StarDist2D.from_pretrained("2D_versatile_fluo")
 
+
     def stardist_fn(tile):
         model = _get_model()
         ...