Merge pull request #197 from Dana-Farber-AIOS/release-1.0.2

Release v1.0.2

README.md

@@ -16,7 +16,7 @@ A toolkit for computational pathology and machine learning.
 
 There are several ways to install `PathML`:
 
-1. pip install (recommended for users)
+1. pip install (**recommended for users**)
 2. clone repo to local machine and install from source (recommended for developers/contributors)
 
 Options (1) and (2) require that you first install all external dependencies:
@@ -32,7 +32,7 @@ Download Miniconda [here](https://docs.conda.io/en/latest/miniconda.html)
 
 Create conda environment
 ````
-conda create --name pathml python=3.8 numpy=1.18.5
+conda create --name pathml python=3.8
 conda activate pathml
 ````
 

docs/source/h5path.rst

@@ -98,6 +98,7 @@ Here we examine the **h5path** file format in detail:
     root/                           (Group)
     ├── fields/                     (Group)
     │   ├── name                    (Attribute, str)
+    │   ├── shape                   (Attribute, tuple)
     │   ├── labels                  (Group)
     │   │   ├── label1              (Attribute, [str, int, float, array])
     │   │   ├── label2              (Attribute, [str, int, float, array])

environment.yml

@@ -7,17 +7,17 @@ channels:
 dependencies:
     - pip==21.2.2
     - python==3.8
-    - numpy>=1.16.4,<1.19.0
-    - scipy>=1.5.4
-    - scikit-image>=0.16.2
+    - numpy==1.19.5
+    - scipy==1.7.1
+    - scikit-image==0.18.3
     - matplotlib==3.1.3
-    - python-spams>=2.6.1
+    - python-spams==2.6.1
     - openjdk==8.0.152
     - pytorch==1.9.0
     - h5py==3.1.0
     - dask==2021.7.1
     - pydicom==2.1.2
-    - pytest>=6.2.2
+    - pytest==6.2.5
     - pre-commit==2.13.0
     - coverage==5.5
     - pip:
@@ -27,12 +27,9 @@ dependencies:
         - python-bioformats==4.0.0
         - scanpy==1.7.2
         - anndata==0.7.6
-        - deepcell>=0.9.0
-        - cellpose>=0.6.1
-        - tensorly==0.6.0
-        - ipython>=7.20.0
-        - sphinx>=3.4.3
-        - nbsphinx>=0.8.1
-        - nbsphinx-link>=1.3.0
-        - sphinx-rtd-theme>=0.5.1
-        - sphinx-autoapi>=1.8.3
+        - ipython==7.27.0
+        - sphinx==4.2.0
+        - nbsphinx==0.8.7
+        - nbsphinx-link==1.3.0
+        - sphinx-rtd-theme==1.0.0
+        - sphinx-autoapi==1.8.4

pathml/core/h5managers.py

@@ -56,6 +56,7 @@ def __init__(self, h5path=None, slidedata=None):
             #    create group
             fieldsgroup = self.h5.create_group("fields")
             fieldsgroup.attrs["name"] = slidedata.name
+            fieldsgroup.attrs["shape"] = slidedata.slide.get_image_shape()
             labelsgroup = self.h5["fields"].create_group("labels")
             if slidedata.labels:
                 for key, label in slidedata.labels.items():
@@ -638,6 +639,6 @@ def check_valid_h5path_format(h5path):
     """
     assert set(h5path.keys()) == {"fields", "array", "masks", "counts", "tiles"}
     assert set(h5path["fields"].keys()) == {"labels", "slide_type"}
-    assert set(h5path["fields"].attrs.keys()) == {"name"}
+    assert set(h5path["fields"].attrs.keys()) == {"name", "shape"}
     # slide_type attributes are not enforced
     return True

pathml/core/slide_data.py

@@ -104,6 +104,7 @@ def __init__(
                 [
                     isinstance(val, (str, np.ndarray))
                     or np.issubdtype(type(val), np.number)
+                    or np.issubdtype(type(val), np.bool_)
                     for val in labels.values()
                 ]
             ), (
@@ -336,7 +337,10 @@ def shape(self):
         Returns:
             Tuple[int, int]: Shape of image (H, W)
         """
-        return self.slide.get_image_shape()
+        if self.backend == "h5path":
+            return tuple(self.h5manager.h5["fields"].attrs["shape"])
+        else:
+            return self.slide.get_image_shape()
 
     @staticmethod
     def _create_tile_dataset(slidedata):

pathml/core/tile.py

@@ -77,6 +77,7 @@ def __init__(
                 [
                     isinstance(val, (str, np.ndarray))
                     or np.issubdtype(type(val), np.number)
+                    or np.issubdtype(type(val), np.bool_)
                     for val in labels.values()
                 ]
             ), (

setup.py

@@ -5,7 +5,7 @@
 
 setuptools.setup(
     name="pathml",
-    version="1.0.1",
+    version="1.0.2",
     author="Jacob Rosenthal, Ryan Carelli et al.",
     author_email="[email protected]",
     description="Tools for computational pathology",
@@ -17,26 +17,24 @@
             "Source Code": "https://github.com/Dana-Farber-AIOS/pathml",
     },
     install_requires=[
-        "pip==21.2.2",
-        "numpy>=1.16.4,<1.19.0",
-        "pandas==1.1.5",
-        "scipy==1.5.4",
+        "pip",
+        "numpy>=1.16.4",
+        "pandas",
+        "scipy",
         "scikit-image",
-        "statsmodels==0.12.2",
-        "matplotlib==3.1.3",
-        "openslide-python==1.1.2",
-        "pydicom==2.1.2",
-        "h5py==3.1.0",
-        "spams==2.6.2.5",
-        "scikit-learn==0.24.2",
-        "dask[distributed]==2021.7.1",
-        "anndata==0.7.6",
-        "scanpy==1.7.2",
-        "pre-commit==2.13.0",
-        "torch==1.9.0",
-        "opencv-contrib-python==4.5.3.56",
-        "tensorly==0.6.0",
-        "python-bioformats==4.0.0",
+        "statsmodels",
+        "matplotlib",
+        "openslide-python",
+        "pydicom",
+        "h5py",
+        "spams",
+        "scikit-learn",
+        "dask[distributed]",
+        "anndata>=0.7.6",
+        "scanpy",
+        "torch",
+        "opencv-contrib-python",
+        "python-bioformats",
     ],
     classifiers=[
         "License :: OSI Approved :: GNU General Public License v2 (GPLv2)",

tests/conftest.py

@@ -39,6 +39,7 @@ def create_HE_tile():
         "test_array_label": np.array([2, 3, 4]),
         "test_int_label": 3,
         "test_float_label": 3.0,
+        "test_bool_label": True,
     }
     tile = Tile(image=im_np_rgb, coords=(1, 3), masks=masks, labels=labs)
     return tile
@@ -83,6 +84,7 @@ def tileVectra():
         "test_array_label": np.array([2, 3, 4]),
         "test_int_label": 3,
         "test_float_label": 3.0,
+        "test_bool_label": True,
     }
 
     tile = Tile(

tests/core_tests/conftest.py

@@ -5,6 +5,7 @@
 
 import anndata
 import numpy as np
+
 # define fixtures here, and use them throughout the other tests in core_tests/
 import pytest
 from pathml.core import SlideData, SlideDataset, Tile, VectraSlide, types
@@ -17,6 +18,7 @@ def example_slide_data():
         "test_array_label": np.array([2, 3, 4]),
         "test_int_label": 3,
         "test_float_label": 3.0,
+        "test_bool_label": True,
     }
     wsi = SlideData(
         "tests/testdata/small_HE.svs", name=f"test", labels=labs, backend="openslide"
@@ -31,6 +33,7 @@ def example_slide_data_with_tiles(tile):
         "test_array_label": np.array([2, 3, 4]),
         "test_int_label": 3,
         "test_float_label": 3.0,
+        "test_bool_label": True,
     }
     adata = anndata.AnnData()
     wsi = SlideData(
@@ -52,6 +55,7 @@ def slide_dataset(example_slide_data_with_tiles):
         "test_array_label": np.array([2, 3, 4]),
         "test_int_label": 3,
         "test_float_label": 3.0,
+        "test_bool_label": True,
     }
     slide_list = [
         SlideData(
@@ -74,6 +78,7 @@ def slide_dataset_with_tiles(tile, example_slide_data_with_tiles):
         "test_array_label": np.array([2, 3, 4]),
         "test_int_label": 3,
         "test_float_label": 3.0,
+        "test_bool_label": True,
     }
     slide_list = [
         SlideData(

tests/core_tests/test_slide_data.py

@@ -126,7 +126,9 @@ def test_read_write_heslide(tmp_path, example_slide_data_with_tiles):
     path = tmp_path / "testhe.h5"
     slidedata.write(path)
     readslidedata = SlideData(path)
+    repr(readslidedata)
     assert readslidedata.name == slidedata.name
+    assert readslidedata.shape == slidedata.shape
     np.testing.assert_equal(readslidedata.labels, slidedata.labels)
     if slidedata.masks is None:
         assert readslidedata.masks is None

tests/integration_tests/test_pipeline_running.py

@@ -83,6 +83,7 @@ def test_pipeline_bioformats_tiff(tmp_path, dist, tile_size):
 @pytest.mark.parametrize("dist", [False, True])
 @pytest.mark.parametrize("tile_size", [1000, (1920, 1440)])
 def test_pipeline_bioformats_vectra(tmp_path, dist, tile_size):
+    SegmentMIF = pytest.importorskip("pathml.preprocessing.transforms.SegmentMIF")
     slide = VectraSlide("tests/testdata/small_vectra.qptiff")
     pipeline = Pipeline(
         [

tests/preprocessing_tests/test_pipeline.py

@@ -7,9 +7,9 @@
 import numpy as np
 import pandas as pd
 import copy
+import pytest
 
 from pathml.preprocessing import Pipeline
-
 from pathml.preprocessing import (
     MedianBlur,
     GaussianBlur,
@@ -22,7 +22,6 @@
     StainNormalizationHE,
     NucleusDetectionHE,
     TissueDetectionHE,
-    SegmentMIF,
     QuantifyMIF,
     CollapseRunsVectra,
 )
@@ -75,6 +74,7 @@ def test_pipeline_mif(tileVectra):
     """
     Run MIF pipeline
     """
+    SegmentMIF = pytest.importorskip("pathml.preprocessing.transforms.SegmentMIF")
     orig_tile = copy.copy(tileVectra)
     pipe = Pipeline(
         [

tests/preprocessing_tests/test_transforms.py

@@ -21,7 +21,6 @@
     LabelArtifactTileHE,
     LabelWhiteSpaceHE,
     QuantifyMIF,
-    SegmentMIF,
     CollapseRunsVectra,
     CollapseRunsCODEX,
 )
@@ -130,6 +129,7 @@ def test_binary_label_transforms(tileHE, transform):
 
 
 def test_segment_mif(tileVectra):
+    SegmentMIF = pytest.importorskip("pathml.preprocessing.transforms.SegmentMIF")
     vectra_collapse = CollapseRunsVectra()
     vectra_collapse.apply(tileVectra)
     t = SegmentMIF(nuclear_channel=0, cytoplasm_channel=1)
@@ -141,6 +141,7 @@ def test_segment_mif(tileVectra):
 
 
 def test_quantify_mif(tileVectra):
+    SegmentMIF = pytest.importorskip("pathml.preprocessing.transforms.SegmentMIF")
     t = QuantifyMIF("cell_segmentation")
     with pytest.raises(AssertionError):
         t.apply(tileVectra)
@@ -178,10 +179,14 @@ def test_collapse_runs_vectra(tileVectra):
         LabelArtifactTileHE(),
         LabelWhiteSpaceHE(),
         QuantifyMIF(segmentation_mask="test"),
-        SegmentMIF(nuclear_channel=0, cytoplasm_channel=1),
         CollapseRunsVectra(),
         CollapseRunsCODEX(z=0),
     ],
 )
 def test_repr(transform):
     repr(transform)
+
+
+def test_segmentMIF_repr():
+    SegmentMIF = pytest.importorskip("pathml.preprocessing.transforms.SegmentMIF")
+    repr(SegmentMIF(nuclear_channel=0, cytoplasm_channel=1))