Merge 5e1ca87 into ef1a4a6

.gitignore

@@ -33,3 +33,4 @@ Pipfile
 Pipfile.lock
 .coverage
 .python-version
+.coverage.*

.readthedocs.yml

@@ -5,6 +5,6 @@ sphinx:
   fail_on_warning: true
 
 python:
-  version: "3.7"
+  version: "3.8"
   install:
     - requirements: docs/rtd-requirements.txt

.travis.yml

@@ -6,10 +6,10 @@ git:
 language: python
 
 python:
-  - 3.7
+  - 3.8
 
-# Install the updated versions of importlib-metadata and setuptools, as Travis 3.7 environments
-# contains a very old version of importlib-metadata which is incompatible for 3.7+
+# Install the updated versions of importlib-metadata and setuptools, as Travis 3.8 environments
+# contains a very old version of importlib-metadata which is incompatible for 3.8+
 # We add python path to enable testing jupyter notebooks
 install:
   - pip install -U importlib_metadata setuptools
@@ -30,7 +30,7 @@ script:
 jobs:
   include:
     - stage: pypi_deploy
-      python: 3.7
+      python: 3.8
       deploy:
         provider: pypi
         user: $PYPI_USERNAME
@@ -39,7 +39,7 @@ jobs:
           tags: true
     - stage: docker_deploy
       if: tag IS present
-      python: 3.7
+      python: 3.8
       script:
         - "travis_wait 120 sleep 7200 &"
         - echo "$DOCKER_PASSWORD" | docker login -u "$DOCKER_USERNAME" --password-stdin

Dockerfile

@@ -1,4 +1,4 @@
-FROM python:3.7
+FROM python:3.8
 
 # system maintenance
 RUN apt-get update

ark/analysis/dimensionality_reduction.py

@@ -3,7 +3,7 @@
 from sklearn.manifold import TSNE
 import matplotlib.pyplot as plt
 import seaborn as sns
-import umap.umap_ as umap
+import umap
 import os
 
 from ark.utils import misc_utils

ark/phenotyping/cell_cluster_utils_test.py

@@ -11,7 +11,7 @@
 from matplotlib.colors import ListedColormap
 import numpy as np
 import pandas as pd
-from skimage.draw import circle
+from skimage.draw import disk
 import skimage.io as io
 import scipy.ndimage as ndimage
 from sklearn.utils import shuffle

ark/phenotyping/pixel_cluster_utils_test.py

@@ -11,7 +11,7 @@
 from matplotlib.colors import ListedColormap
 import numpy as np
 import pandas as pd
-from skimage.draw import circle
+from skimage.draw import disk
 import skimage.io as io
 import scipy.ndimage as ndimage
 from sklearn.utils import shuffle
@@ -519,8 +519,8 @@ def test_filter_with_nuclear_mask(sub_dir, exclude, capsys):
             rand_img = np.random.randint(1, 16, size=(1, 10, 10))
 
             # draw a dummy nucleus and store the coords
-            nuclear_x, nuclear_y = circle(
-                base_center[0] + offset, base_center[1] + offset, base_radius
+            nuclear_x, nuclear_y = disk(
+                (base_center[0] + offset, base_center[1] + offset), base_radius
             )
             rand_img[0, nuclear_x, nuclear_y] = 0
             nuclear_coords[fov] = (nuclear_x, nuclear_y)

ark/segmentation/regionprops_extraction.py

@@ -124,7 +124,7 @@ def num_concavities(prop, **kwargs):
     diff_img = convex_image ^ cell_image
 
     if np.sum(diff_img) > 0:
-        labeled_diff_img = label(diff_img)
+        labeled_diff_img = label(diff_img, connectivity=1)
         hull_prop_df = pd.DataFrame(regionprops_table(labeled_diff_img,
                                                       properties=['area', 'perimeter']))
         hull_prop_df['compactness'] = np.square(hull_prop_df['perimeter']) / hull_prop_df['area']

ark/segmentation/regionprops_extraction_test.py

@@ -20,7 +20,8 @@ def test_major_minor_axis_ratio():
     prop_info = regionprops(sample_arr)[0]
 
     major_minor_rat = regionprops_extraction.major_minor_axis_ratio(prop_info)
-    assert np.round(major_minor_rat, 4) == 1.1524
+    desired_value = 1.1524
+    np.testing.assert_allclose(actual=major_minor_rat, desired=desired_value, rtol=0.1, atol=1e-5)
 
     class Regionprop(object):
         pass
@@ -42,7 +43,8 @@ def test_perim_square_over_area():
     prop_info = regionprops(sample_arr)[0]
 
     perim_area_rat = regionprops_extraction.perim_square_over_area(prop_info)
-    assert np.round(perim_area_rat, 4) == 39.3630
+    desired_value = 39.3630
+    np.testing.assert_allclose(actual=perim_area_rat, desired=desired_value, rtol=0.1, atol=1e-5)
 
 
 def test_major_axis_equiv_diam_ratio():
@@ -54,7 +56,8 @@ def test_major_axis_equiv_diam_ratio():
     prop_info = regionprops(sample_arr)[0]
 
     major_diam_rat = regionprops_extraction.major_axis_equiv_diam_ratio(prop_info)
-    assert np.round(major_diam_rat, 4) == 1.7664
+    desired_value = 1.7664
+    np.testing.assert_allclose(actual=major_diam_rat, desired=desired_value, rtol=0.1, atol=1e-5)
 
 
 def test_convex_hull_resid():
@@ -66,7 +69,8 @@ def test_convex_hull_resid():
     prop_info = regionprops(sample_arr)[0]
 
     convex_res = regionprops_extraction.convex_hull_resid(prop_info)
-    assert np.round(convex_res, 4) == 0.6605
+    desired_value = 0.6605
+    np.testing.assert_allclose(actual=convex_res, desired=desired_value, rtol=0.1, atol=1e-5)
 
 
 def test_nc_ratio():
@@ -115,7 +119,8 @@ def test_centroid_dif():
     prop_info = regionprops(sample_arr)[0]
 
     centroid_dist = regionprops_extraction.centroid_dif(prop_info)
-    assert np.round(centroid_dist, 4) == 0.3562
+    desired_value = 0.3562
+    np.testing.assert_allclose(actual=centroid_dist, desired=desired_value, rtol=0.1, atol=1e-5)
 
 
 def test_num_concavities():

ark/utils/deepcell_service_utils.py

@@ -5,15 +5,16 @@
 import requests
 from requests.adapters import HTTPAdapter
 from requests.exceptions import RetryError
-from requests.packages.urllib3.util.retry import Retry
+from requests.packages.urllib3.util import Retry
 import time
 from tqdm.notebook import tqdm
 from urllib.parse import unquote_plus
 import warnings
 from concurrent.futures import ThreadPoolExecutor
 import numpy as np
 from scipy import stats
-from skimage import io, external
+from skimage import io
+from tifffile import imread
 from io import BytesIO
 from ark.utils import misc_utils
 from zipfile import ZipFile, ZIP_DEFLATED
@@ -323,7 +324,7 @@ def _convert_deepcell_seg_masks(seg_mask: bytes) -> np.ndarray:
         np.ndarray: The segmentation masks, converted from floating point 64-bit to integer
         16-bit via `scipy.stats.rankdata`
     """
-    float_mask = external.tifffile.imread(BytesIO(seg_mask))
+    float_mask = imread(BytesIO(seg_mask))
 
     # Reshape as ranked_mask returns a 1D numpy array, dims:  n^2 x 1 -> 1 x n x n
     shape = float_mask.shape

ark/utils/plot_utils_test.py

@@ -9,7 +9,7 @@
 import pytest
 
 from ark.utils import plot_utils, test_utils
-from skimage.draw import circle
+from skimage.draw import disk
 
 from pathlib import Path
 import natsort
@@ -23,7 +23,7 @@ def _generate_segmentation_labels(img_dims, num_cells=20):
 
     for i in range(num_cells):
         r, c = np.random.randint(radius, img_dims[0] - radius, 2)
-        rr, cc = circle(r, c, radius)
+        rr, cc = disk((r, c), radius)
         labels[rr, cc] = i
 
     return labels

ark/utils/synthetic_spatial_datagen.py

@@ -2,7 +2,7 @@
 import xarray as xr
 
 from copy import deepcopy
-from skimage.draw import circle
+from skimage.draw import disk
 
 
 def generate_test_dist_matrix(num_A=100, num_B=100, num_C=100,
@@ -263,10 +263,9 @@ def generate_two_cell_seg_mask(size_img=(1024, 1024), cell_radius=10):
     center_2 = (size_img[0] // 2, size_img[0] // 2 + cell_radius * 2 - 1)
 
     # generate the coordinates of each nuclear disk
-    cell_region_1_x, cell_region_1_y = circle(center_1[0], center_1[1], cell_radius,
-                                              shape=size_img)
-    cell_region_2_x, cell_region_2_y = circle(center_2[0], center_2[1], cell_radius,
-                                              shape=size_img)
+    cell_region_1_x, cell_region_1_y = disk(center=center_1, radius=cell_radius, shape=size_img)
+
+    cell_region_2_x, cell_region_2_y = disk(center=center_2, radius=cell_radius, shape=size_img)
 
     # assign the respective cells value according to their label
     sample_segmentation_mask[cell_region_1_x, cell_region_1_y] = 1
@@ -314,8 +313,9 @@ def generate_two_cell_nuc_signal(segmentation_mask, cell_centers,
         center = cell_centers[cell]
 
         # generate nuclear region
-        nuc_region_x, nuc_region_y = circle(center[0], center[1],
-                                            nuc_radius + nuc_uncertainty_length, shape=size_img)
+        nuc_region_x, nuc_region_y = disk(center=center,
+                                          radius=nuc_radius + nuc_uncertainty_length,
+                                          shape=size_img)
 
         # set nuclear signal
         sample_nuclear_signal[nuc_region_x, nuc_region_y] = nuc_signal_strength
@@ -364,14 +364,14 @@ def generate_two_cell_memb_signal(segmentation_mask, cell_centers,
         center = cell_centers[cell]
 
         # generate coordinates of the cell region
-        cell_region_x, cell_region_y = circle(center[0], center[1],
-                                              cell_radius + memb_uncertainty_length,
-                                              shape=size_img)
+        cell_region_x, cell_region_y = disk(center=center,
+                                            radius=cell_radius + memb_uncertainty_length,
+                                            shape=size_img)
 
         # generate coordinates of the non-membrane region
-        non_memb_region_x, non_memb_region_y = circle(center[0], center[1],
-                                                      cell_radius - memb_thickness,
-                                                      shape=size_img)
+        non_memb_region_x, non_memb_region_y = disk(center=center,
+                                                    radius=cell_radius - memb_thickness,
+                                                    shape=size_img)
 
         # perform circle subtraction to generate membrane region
         sample_membrane_signal[cell_region_x, cell_region_y] = memb_signal_strength

ark/utils/tiff_utils.py

@@ -1,6 +1,6 @@
 from fractions import Fraction
 import numpy as np
-from skimage.external.tifffile import TiffFile, TiffWriter
+from tifffile import TiffFile, TiffWriter
 import json
 import datetime
 from itertools import compress
@@ -33,7 +33,8 @@ def read_mibitiff(file, channels=None):
 
             # get tags as json
             description = json.loads(
-                page.tags['image_description'].value.decode('utf-8')
+                # `ImageDescription` tag code is 270
+                page.tags[270].value
             )
 
             # only load supplied channels
@@ -69,8 +70,8 @@ def _check_version(file):
     Raises:
         ValueError
     """
-    filetype = file.pages[0].tags.get('software')
-    if not (filetype and filetype.value.decode('utf-8').startswith('IonpathMIBI')):
+    filetype = file.pages[0].tags.get('Software')
+    if not (filetype and filetype.value.startswith('IonpathMIBI')):
         raise ValueError('File is not of type IonpathMIBI...')
 
 
@@ -106,16 +107,22 @@ def write_mibitiff(filepath, img_data, channel_tuples, metadata):
         (286, '2i', 1, _micron_to_cm(metadata['coordinates'][0])),
         (287, '2i', 1, _micron_to_cm(metadata['coordinates'][1]))
     ]
-    resolution = (img_data.shape[0] * 1e4 / float(metadata['size']),
-                  img_data.shape[1] * 1e4 / float(metadata['size']),
-                  'cm')
+    _resolution = (img_data.shape[0] * 1e4 / float(metadata['size']),
+                   img_data.shape[1] * 1e4 / float(metadata['size']))
+
+    # resolutionunit / RESUNIT = 3 is "CENTIMETER"
+    _resolutionunit = 3
+
+    # Compression uses ZLIB, with level 6
+    _compression = "zlib"
+    _compressionargs = {"level": 6}
 
     description = {}
     for key, value in metadata.items():
         if key in _PREFIXED_METADATA_ATTRIBUTES:
             description[f'mibi.{key}'] = value
 
-    with TiffWriter(filepath, software="IonpathMIBIv1.0") as infile:
+    with TiffWriter(filepath) as infile:
         for index, channel_tuple in enumerate(channel_tuples):
             mass, target = channel_tuple
             _metadata = description.copy()
@@ -132,9 +139,15 @@ def write_mibitiff(filepath, img_data, channel_tuples, metadata):
             max_value = (341, range_dtype, 1, ranges[index][1])
             page_tags = coordinates + [page_name, min_value, max_value]
 
-            infile.save(
-                img_data[:, :, index], compress=6, resolution=resolution,
-                extratags=page_tags, metadata=_metadata,
+            infile.write(
+                data=img_data[:, :, index],
+                compression=_compression,
+                compressionargs=_compressionargs,
+                resolution=_resolution,
+                resolutionunit=_resolutionunit,
+                software="IonpathMIBIv1.0",
+                extratags=page_tags,
+                metadata=_metadata,
                 datetime=datetime.datetime.strptime(metadata['date'], '%Y-%m-%dT%H:%M:%S')
             )
 

docs/_rtd/development.md

@@ -2,7 +2,7 @@
 
 If you wish to do higher-level development on top of `ark`, we recommend setting up a virtual environment. We highly recommend using `conda` virtual environments. To be able to set one up, you will need to install the Anaconda package.
 
-### Setting up Your Virtual Environment
+### Setting up Your Virtual Environment - Anaconda
 
 #### Installing Anaconda
 
@@ -11,7 +11,8 @@ For a step-by-step guide of how to install Anaconda, please refer to these links
 * https://github.com/conda-forge/miniforge/releases for Mac (arm64 / Apple Silicon) users
 * https://docs.anaconda.com/anaconda/install/windows/ for Windows users
 
-##### Notes for Mac users
+
+**Notes for Mac users**
 
 We recommend following the command line installer instructions as users have reported recent issues with the graphical installer. 
 
@@ -43,23 +44,25 @@ Miniforge contains conda with native Apple Silicon support. There are a few inst
 
 Now that Anaconda is installed, you can now create a `conda` environment. 
 
-To do so, on your command line, type `conda create -n <my_env> python=3.7`, where `<my_env>` is a name you set. Our codebase only supports Python 3.7, so please do not change the `python=3.7` flag when creating your environment. 
+To do so, on your command line, type `conda create -n <my_env> python=3.8`, where `<my_env>` is a name you set. Our codebase currently supports up to Python 3.8.
 
 Say yes to any prompts and your `conda` environment will be created! 
 
 To verify installation, activate your `conda` environment with `conda activate <my_env>`. If you see `(<my_env>)` on the far left of the command prompt, you have successfully created and activated your environment. Type `conda deactivate` to exit at any time.
 
 **Apple Silicon Installation**
 
+`Ark` now supports development with Apple Silicon.
+
 Currently there is not a native M1 implementation of ark-analysis for development, so it will need to go through Rosetta 2 (the Intel to Arm transition layer). Luckily, this isn't something you'll have to deal with, as `conda` makes it straightforward.
 
-1. Create a Python 3.6 `conda` environment called `my_env` using osx-64 (compiled binaries for Intel Macs as the default installation type)
+1. Create a Python 3.8 `conda` environment called `my_env`.
     ```sh
-    CONDA_SUBDIR=osx-64 conda create -n <my_env> python=3.6
+    conda create -n <my_env> python=3.8
     ```
 
-2. Test to make sure the `platform.machine()` function reports `x86_64` in the terminal.
-    ```
+2. Test to make sure the `platform.machine()` function reports `arm64` in the terminal.
+    ```sh
     conda activate <my_env>
     python -c "import platform;print(platform.machine())"
     ```
@@ -70,7 +73,7 @@ Currently there is not a native M1 implementation of ark-analysis for developmen
     ```
 4. The prompt may ask you to deactivate and reactivate the environment as well.
 
-Now any package that is installed in `my_env` will targeted for `x86_64`.
+Now any package that is installed in `my_env` will targeted for `arm64`.
 
 ### Setting up ark-analysis for development