Replace most instances of starfish.image.Filter.Reduce with imagestac…

…k.reduce (#1548) Unless the reduction is happening multiple times, it's simpler to represent a stack reduction as imagestack.reduce Test plan: travis thanks!
spacetx · Oct 10, 2019 · 3cd0df6 · 3cd0df6
1 parent 6bf1d96
commit 3cd0df6
Show file tree

Hide file tree

Showing 13 changed files with 43 additions and 77 deletions.
diff --git a/docs/source/_static/data_processing_examples/iss_pipeline.py b/docs/source/_static/data_processing_examples/iss_pipeline.py
@@ -19,9 +19,8 @@ def iss_pipeline(fov, codebook):
     # register the raw image
     learn_translation = LearnTransform.Translation(reference_stack=fov.get_image('dots'),
                                                    axes=Axes.ROUND, upsampling=100)
-    max_projector = Filter.Reduce(
-        (Axes.CH, Axes.ZPLANE), func="max", module=FunctionSource.np)
-    transforms_list = learn_translation.run(max_projector.run(primary_image))
+    transforms_list = learn_translation.run(
+        primary_image.reduce({Axes.CH, Axes.ZPLANE}, func="max"))
     warp = ApplyTransform.Warp()
     registered = warp.run(primary_image, transforms_list=transforms_list,  in_place=False, verbose=True)
 
@@ -39,8 +38,7 @@ def iss_pipeline(fov, codebook):
     )
 
     # detect spots using laplacian of gaussians approach
-    dots_max_projector = Filter.Reduce((Axes.ROUND, Axes.ZPLANE), func="max", module=FunctionSource.np)
-    dots_max = dots_max_projector.run(fov.get_image('dots'))
+    dots_max = fov.get_image('dots').reduce((Axes.ROUND, Axes.ZPLANE), func="max", module=FunctionSource.np)
     # locate spots in a reference image
     spots = bd.run(reference_image=dots_max, image_stack=filtered)
 

diff --git a/docs/source/_static/tutorials/exec_image_manipulations.py b/docs/source/_static/tutorials/exec_image_manipulations.py
@@ -73,8 +73,7 @@
 #
 from starfish.image import Filter
 
-max_projector = Filter.Reduce((Axes.CH,), func="max", module=FunctionSource.np)
-projected_image: starfish.ImageStack = max_projector.run(image)
+projected_image: starfish.ImageStack = image.reduce({Axes.CH}, func="max")
 
 ###################################################################################################
 # To demonstrate the effect, the below figure displays each channel of round :code:`1` in the
@@ -108,4 +107,4 @@
 ax6.set_axis_off()
 
 # fix matplotlib whitespace
-f.tight_layout()
+f.tight_layout()
diff --git a/docs/source/getting_started/loading_data/index.rst b/docs/source/getting_started/loading_data/index.rst
@@ -108,9 +108,8 @@ we'll collapse all the spots across channels in each round, mimicing a "dots" im
 
     In[11]: from starfish.image import Filter
     In[12]: from starfish.types import FunctionSource
-    In[13]: max_projector = Filter.Reduce((Axes.CH,), func="max", module=FunctionSource.np)
-    In[14]: max_projector.run(image)
-    Out[14]: <starfish.ImageStack (r: 4, c: 1, z: 1, y: 1000, x: 1000)>
+    In[13]: image.reduce({Axes.CH}, func="max")
+    Out[13]: <starfish.ImageStack (r: 4, c: 1, z: 1, y: 1000, x: 1000)>
 
 Visualizing Data
 ----------------
@@ -125,10 +124,10 @@ to enable the :code:`qt` environment in IPython:
 
 .. code-block:: python
 
-    In[15]: ipython = get_ipython()
-    In[16]: ipython.magic("gui qt5")
-    In[17]: starfish.display(image)
-    Out[17]: <napari.components._viewer.model.Viewer at 0x15f7b44e0>
+    In[14]: ipython = get_ipython()
+    In[15]: ipython.magic("gui qt5")
+    In[16]: starfish.display(image)
+    Out[16]: <napari.components._viewer.model.Viewer at 0x15f7b44e0>
 
 Typing the above code should display an image viewer that looks something like this:,
 

diff --git a/notebooks/ISS.ipynb b/notebooks/ISS.ipynb
@@ -137,13 +137,8 @@
    "source": [
     "from starfish.image import Filter\n",
     "\n",
-    "rcz_max_projector = Filter.Reduce(\n",
-    "    (Axes.ROUND, Axes.CH, Axes.ZPLANE,), func=\"max\", module=FunctionSource.np)\n",
-    "per_round_max_projector = Filter.Reduce(\n",
-    "    (Axes.CH, Axes.ZPLANE,), func=\"max\", module=FunctionSource.np)\n",
-    "\n",
     "dots = fov.get_image(\"dots\")\n",
-    "dots_single_plane = rcz_max_projector.run(dots)\n",
+    "dots_single_plane = dots.reduce({Axes.ROUND, Axes.CH, Axes.ZPLANE}, func=\"max\")\n",
     "imshow_plane(dots_single_plane, title=\"Anchor channel, all RNA molecules\")"
    ]
   },
@@ -161,7 +156,7 @@
    "outputs": [],
    "source": [
     "nuclei = fov.get_image(\"nuclei\")\n",
-    "nuclei_single_plane = rcz_max_projector.run(nuclei)\n",
+    "nuclei_single_plane = nuclei.reduce({Axes.ROUND, Axes.CH, Axes.ZPLANE}, func=\"max\")\n",
     "imshow_plane(nuclei_single_plane, title=\"Nuclei (DAPI) channel\")"
    ]
   },
@@ -180,8 +175,6 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "from starfish.image import Filter\n",
-    "\n",
     "# filter raw data\n",
     "masking_radius = 15\n",
     "filt = Filter.WhiteTophat(masking_radius, is_volume=False)\n",
@@ -235,7 +228,7 @@
     "from starfish.image import ApplyTransform, LearnTransform\n",
     "\n",
     "learn_translation = LearnTransform.Translation(reference_stack=dots, axes=Axes.ROUND, upsampling=1000)\n",
-    "transforms_list = learn_translation.run(per_round_max_projector.run(imgs))\n",
+    "transforms_list = learn_translation.run(imgs.reduce({Axes.CH, Axes.ZPLANE}, func=\"max\"))\n",
     "warp = ApplyTransform.Warp()\n",
     "registered_imgs = warp.run(filtered_imgs, transforms_list=transforms_list, in_place=False, verbose=True)"
    ]
@@ -266,8 +259,7 @@
     "    measurement_type='mean',\n",
     ")\n",
     "\n",
-    "dots_max_projector = Filter.Reduce((Axes.ROUND, Axes.ZPLANE), func=\"max\", module=FunctionSource.np)\n",
-    "dots_max = dots_max_projector.run(dots)\n",
+    "dots_max = dots.reduce((Axes.ROUND, Axes.ZPLANE), func=\"max\", module=FunctionSource.np)\n",
     "spots = bd.run(image_stack=registered_imgs, reference_image=dots_max)\n",
     "\n",
     "decoder = DecodeSpots.PerRoundMaxChannel(codebook=experiment.codebook)\n",
@@ -300,10 +292,10 @@
     "stain_thresh = .22  # binary mask for overall cells // binarization of stain\n",
     "min_dist = 57\n",
     "\n",
-    "registered_mp = per_round_max_projector.run(registered_imgs).xarray.squeeze()\n",
+    "registered_mp = registered_imgs.reduce({Axes.CH, Axes.ZPLANE}, func=\"max\").xarray.squeeze()\n",
     "stain = np.mean(registered_mp, axis=0)\n",
     "stain = stain/stain.max()\n",
-    "nuclei = rcz_max_projector.run(nuclei)\n",
+    "nuclei = nuclei.reduce({Axes.ROUND, Axes.CH, Axes.ZPLANE}, func=\"max\")\n",
     "\n",
     "\n",
     "seg = Segment.Watershed(\n",
@@ -363,9 +355,9 @@
     "GENE2 = 'VIM'\n",
     "\n",
     "rgb = np.zeros(registered_imgs.tile_shape + (3,))\n",
-    "nuclei_numpy = rcz_max_projector.run(nuclei)._squeezed_numpy(Axes.ROUND, Axes.CH, Axes.ZPLANE)\n",
+    "nuclei_numpy = nuclei.reduce({Axes.ROUND, Axes.CH, Axes.ZPLANE}, func=\"max\")._squeezed_numpy(Axes.ROUND, Axes.CH, Axes.ZPLANE)\n",
     "rgb[:,:,0] = nuclei_numpy\n",
-    "dots_numpy = rcz_max_projector.run(dots)._squeezed_numpy(Axes.ROUND, Axes.CH, Axes.ZPLANE)\n",
+    "dots_numpy = dots.reduce({Axes.ROUND, Axes.CH, Axes.ZPLANE}, func=\"max\")._squeezed_numpy(Axes.ROUND, Axes.CH, Axes.ZPLANE)\n",
     "rgb[:,:,1] = dots_numpy\n",
     "do = rgb2gray(rgb)\n",
     "do = do/(do.max())\n",

diff --git a/notebooks/MERFISH.ipynb b/notebooks/MERFISH.ipynb
@@ -324,9 +324,6 @@
     "from scipy.stats import scoreatpercentile\n",
     "import warnings\n",
     "\n",
-    "all_max_projector = Filter.Reduce(\n",
-    "    (Axes.ROUND, Axes.CH, Axes.ZPLANE,), func=\"max\", module=FunctionSource.np)\n",
-    "\n",
     "f, (ax1, ax2) = plt.subplots(1, 2, figsize=(10, 5))\n",
     "\n",
     "with warnings.catch_warnings():\n",
@@ -337,7 +334,7 @@
     "    show_image(np.squeeze(prop_results.decoded_image)*(mask > 2), cmap='nipy_spectral', ax=ax1)\n",
     "    ax1.axes.set_axis_off()\n",
     "\n",
-    "    mp_numpy = all_max_projector.run(filtered_imgs)._squeezed_numpy(\n",
+    "    mp_numpy = filtered_imgs.reduce({Axes.ROUND, Axes.CH, Axes.ZPLANE}, func=\"max\")._squeezed_numpy(\n",
     "        Axes.ROUND, Axes.CH, Axes.ZPLANE)\n",
     "    clim = scoreatpercentile(mp_numpy, [0.5, 99.5])\n",
     "    show_image(mp_numpy, clim=clim, ax=ax2)\n",

diff --git a/notebooks/STARmap.ipynb b/notebooks/STARmap.ipynb
@@ -78,8 +78,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "ch_r_max_projector = starfish.image.Filter.Reduce((Axes.CH, Axes.ROUND), func=\"max\")\n",
-    "ch_r_max_projection = ch_r_max_projector.run(stack)\n",
+    "ch_r_max_projection = stack.reduce({Axes.CH, Axes.ROUND}, func=\"max\")\n",
     "\n",
     "f = plt.figure(dpi=150)\n",
     "imshow_plane(ch_r_max_projection, sel={Axes.ZPLANE: 15})"
@@ -134,8 +133,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "ch_z_max_projector = starfish.image.Filter.Reduce((Axes.CH, Axes.ZPLANE), func=\"max\")\n",
-    "projection = ch_r_max_projector.run(stack)\n",
+    "projection = stack.reduce({Axes.CH, Axes.ZPLANE}, func=\"max\")\n",
     "reference_image = projection.sel({Axes.ROUND: 0})\n",
     "\n",
     "ltt = starfish.image.LearnTransform.Translation(\n",
@@ -195,7 +193,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "post_projection = ch_z_max_projector.run(stack)"
+    "post_projection = stack.reduce({Axes.CH, Axes.ZPLANE}, func=\"max\")"
    ]
   },
   {
@@ -374,4 +372,4 @@
  },
  "nbformat": 4,
  "nbformat_minor": 2
-}
+}
diff --git a/notebooks/osmFISH.ipynb b/notebooks/osmFISH.ipynb
@@ -108,8 +108,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "z_max_projector = Filter.Reduce((Axes.ZPLANE,))\n",
-    "mp = z_max_projector.run(imgs_ghp_laplace)"
+    "mp = imgs_ghp_laplace.reduce({Axes.ZPLANE}, func=\"max\")"
    ]
   },
   {
@@ -125,7 +124,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "single_plane = imgs.max_proj(Axes.ZPLANE).xarray.sel({Axes.CH:0}).squeeze()\n",
+    "single_plane = imgs.reduce({Axes.ZPLANE}, func=\"max\").xarray.sel({Axes.CH:0}).squeeze()\n",
     "single_plane_filtered = mp.xarray.sel({Axes.CH: 0}).squeeze()\n",
     "\n",
     "plt.figure(figsize=(10,10))\n",

diff --git a/notebooks/py/ISS.py b/notebooks/py/ISS.py
@@ -90,13 +90,8 @@
 # EPY: START code
 from starfish.image import Filter
 
-rcz_max_projector = Filter.Reduce(
-    (Axes.ROUND, Axes.CH, Axes.ZPLANE,), func="max", module=FunctionSource.np)
-per_round_max_projector = Filter.Reduce(
-    (Axes.CH, Axes.ZPLANE,), func="max", module=FunctionSource.np)
-
 dots = fov.get_image("dots")
-dots_single_plane = rcz_max_projector.run(dots)
+dots_single_plane = dots.reduce({Axes.ROUND, Axes.CH, Axes.ZPLANE}, func="max")
 imshow_plane(dots_single_plane, title="Anchor channel, all RNA molecules")
 # EPY: END code
 
@@ -106,7 +101,7 @@
 
 # EPY: START code
 nuclei = fov.get_image("nuclei")
-nuclei_single_plane = rcz_max_projector.run(nuclei)
+nuclei_single_plane = nuclei.reduce({Axes.ROUND, Axes.CH, Axes.ZPLANE}, func="max")
 imshow_plane(nuclei_single_plane, title="Nuclei (DAPI) channel")
 # EPY: END code
 
@@ -117,8 +112,6 @@
 # EPY: END markdown
 
 # EPY: START code
-from starfish.image import Filter
-
 # filter raw data
 masking_radius = 15
 filt = Filter.WhiteTophat(masking_radius, is_volume=False)
@@ -156,7 +149,7 @@
 from starfish.image import ApplyTransform, LearnTransform
 
 learn_translation = LearnTransform.Translation(reference_stack=dots, axes=Axes.ROUND, upsampling=1000)
-transforms_list = learn_translation.run(per_round_max_projector.run(imgs))
+transforms_list = learn_translation.run(imgs.reduce({Axes.CH, Axes.ZPLANE}, func="max"))
 warp = ApplyTransform.Warp()
 registered_imgs = warp.run(filtered_imgs, transforms_list=transforms_list, in_place=False, verbose=True)
 # EPY: END code
@@ -179,8 +172,7 @@
     measurement_type='mean',
 )
 
-dots_max_projector = Filter.Reduce((Axes.ROUND, Axes.ZPLANE), func="max", module=FunctionSource.np)
-dots_max = dots_max_projector.run(dots)
+dots_max = dots.reduce((Axes.ROUND, Axes.ZPLANE), func="max", module=FunctionSource.np)
 spots = bd.run(image_stack=registered_imgs, reference_image=dots_max)
 
 decoder = DecodeSpots.PerRoundMaxChannel(codebook=experiment.codebook)
@@ -205,10 +197,10 @@
 stain_thresh = .22  # binary mask for overall cells // binarization of stain
 min_dist = 57
 
-registered_mp = per_round_max_projector.run(registered_imgs).xarray.squeeze()
+registered_mp = registered_imgs.reduce({Axes.CH, Axes.ZPLANE}, func="max").xarray.squeeze()
 stain = np.mean(registered_mp, axis=0)
 stain = stain/stain.max()
-nuclei = rcz_max_projector.run(nuclei)
+nuclei = nuclei.reduce({Axes.ROUND, Axes.CH, Axes.ZPLANE}, func="max")
 
 
 seg = Segment.Watershed(
@@ -249,9 +241,9 @@
 GENE2 = 'VIM'
 
 rgb = np.zeros(registered_imgs.tile_shape + (3,))
-nuclei_numpy = rcz_max_projector.run(nuclei)._squeezed_numpy(Axes.ROUND, Axes.CH, Axes.ZPLANE)
+nuclei_numpy = nuclei.reduce({Axes.ROUND, Axes.CH, Axes.ZPLANE}, func="max")._squeezed_numpy(Axes.ROUND, Axes.CH, Axes.ZPLANE)
 rgb[:,:,0] = nuclei_numpy
-dots_numpy = rcz_max_projector.run(dots)._squeezed_numpy(Axes.ROUND, Axes.CH, Axes.ZPLANE)
+dots_numpy = dots.reduce({Axes.ROUND, Axes.CH, Axes.ZPLANE}, func="max")._squeezed_numpy(Axes.ROUND, Axes.CH, Axes.ZPLANE)
 rgb[:,:,1] = dots_numpy
 do = rgb2gray(rgb)
 do = do/(do.max())

diff --git a/notebooks/py/MERFISH.py b/notebooks/py/MERFISH.py
@@ -223,9 +223,6 @@
 from scipy.stats import scoreatpercentile
 import warnings
 
-all_max_projector = Filter.Reduce(
-    (Axes.ROUND, Axes.CH, Axes.ZPLANE,), func="max", module=FunctionSource.np)
-
 f, (ax1, ax2) = plt.subplots(1, 2, figsize=(10, 5))
 
 with warnings.catch_warnings():
@@ -236,7 +233,7 @@
     show_image(np.squeeze(prop_results.decoded_image)*(mask > 2), cmap='nipy_spectral', ax=ax1)
     ax1.axes.set_axis_off()
 
-    mp_numpy = all_max_projector.run(filtered_imgs)._squeezed_numpy(
+    mp_numpy = filtered_imgs.reduce({Axes.ROUND, Axes.CH, Axes.ZPLANE}, func="max")._squeezed_numpy(
         Axes.ROUND, Axes.CH, Axes.ZPLANE)
     clim = scoreatpercentile(mp_numpy, [0.5, 99.5])
     show_image(mp_numpy, clim=clim, ax=ax2)

diff --git a/notebooks/py/STARmap.py b/notebooks/py/STARmap.py
@@ -55,8 +55,7 @@
 # EPY: END code
 
 # EPY: START code
-ch_r_max_projector = starfish.image.Filter.Reduce((Axes.CH, Axes.ROUND), func="max")
-ch_r_max_projection = ch_r_max_projector.run(stack)
+ch_r_max_projection = stack.reduce({Axes.CH, Axes.ROUND}, func="max")
 
 f = plt.figure(dpi=150)
 imshow_plane(ch_r_max_projection, sel={Axes.ZPLANE: 15})
@@ -95,8 +94,7 @@
 # EPY: END markdown
 
 # EPY: START code
-ch_z_max_projector = starfish.image.Filter.Reduce((Axes.CH, Axes.ZPLANE), func="max")
-projection = ch_r_max_projector.run(stack)
+projection = stack.reduce({Axes.CH, Axes.ZPLANE}, func="max")
 reference_image = projection.sel({Axes.ROUND: 0})
 
 ltt = starfish.image.LearnTransform.Translation(
@@ -132,7 +130,7 @@
 # EPY: END markdown
 
 # EPY: START code
-post_projection = ch_z_max_projector.run(stack)
+post_projection = stack.reduce({Axes.CH, Axes.ZPLANE}, func="max")
 # EPY: END code
 
 # EPY: START code

diff --git a/notebooks/py/osmFISH.py b/notebooks/py/osmFISH.py
@@ -68,16 +68,15 @@
 # EPY: END markdown
 
 # EPY: START code
-z_max_projector = Filter.Reduce((Axes.ZPLANE,))
-mp = z_max_projector.run(imgs_ghp_laplace)
+mp = imgs_ghp_laplace.reduce({Axes.ZPLANE}, func="max")
 # EPY: END code
 
 # EPY: START markdown
 #We can now visualize our data before and after filtering
 # EPY: END markdown
 
 # EPY: START code
-single_plane = imgs.max_proj(Axes.ZPLANE).xarray.sel({Axes.CH:0}).squeeze()
+single_plane = imgs.reduce({Axes.ZPLANE}, func="max").xarray.sel({Axes.CH:0}).squeeze()
 single_plane_filtered = mp.xarray.sel({Axes.CH: 0}).squeeze()
 
 plt.figure(figsize=(10,10))

diff --git a/workflows/wdl/iss_published/recipe.py b/workflows/wdl/iss_published/recipe.py
@@ -36,8 +36,7 @@ def process_fov(field_num: int, experiment_str: str):
 
     print("Learning Transform")
     learn_translation = LearnTransform.Translation(reference_stack=dots, axes=Axes.ROUND, upsampling=1000)
-    max_projector = Filter.Reduce((Axes.CH, Axes.ZPLANE))
-    transforms_list = learn_translation.run(max_projector.run(imgs))
+    transforms_list = learn_translation.run(imgs.reduce({Axes.CH, Axes.ZPLANE}, func="max"))
 
     print("Applying transform")
     warp = ApplyTransform.Warp()

diff --git a/workflows/wdl/iss_spaceTX/recipe.py b/workflows/wdl/iss_spaceTX/recipe.py
@@ -24,8 +24,7 @@ def process_fov(field_num: int, experiment_str: str):
 
     fov = experiment[fov_str]
     imgs = fov.get_image(starfish.FieldOfView.PRIMARY_IMAGES)
-    max_projector = starfish.image.Filter.Reduce((Axes.CH, Axes.ZPLANE))
-    dots = max_projector.run(imgs)
+    dots = imgs.reduce({Axes.CH, Axes.ZPLANE}, func="max")
 
     # filter
     filt = starfish.image.Filter.WhiteTophat(masking_radius=15, is_volume=False)
@@ -53,4 +52,4 @@ def process_fov(field_num: int, experiment_str: str):
 
     # save results
     df = decoded.to_decoded_dataframe()
-    return df
+    return df