| Previous changeset 1:41e8efe8df43 (2022-03-11) |
|
Commit message:
planemo upload for repository https://github.com/goeckslab/tools-mti/tree/main/tools/s3segmenter commit 0f4f17235c5961c2fd3d4c30180507f66214c11d |
|
modified:
macros.xml s3segmenter.xml |
|
added:
test-data/stack_probabilities.tiff test-data/test.ome.tiff |
|
removed:
S3segmenter.py rowit.py save_tifffile_pyramid.py |
| b |
| diff -r 41e8efe8df43 -r 96d0d969ebc9 S3segmenter.py --- a/S3segmenter.py Fri Mar 11 23:37:49 2022 +0000 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 |
| [ |
| b"@@ -1,592 +0,0 @@\n-import matplotlib.pyplot as plt\n-import tifffile\n-import os\n-import numpy as np\n-from skimage import io as skio\n-from scipy.ndimage import *\n-import scipy.ndimage as ndi\n-from skimage.morphology import *\n-from skimage.morphology import extrema\n-from skimage import morphology\n-from skimage.measure import regionprops\n-from skimage.transform import resize\n-from skimage.filters import gaussian, threshold_otsu, threshold_local\n-from skimage.feature import peak_local_max\n-from skimage.color import label2rgb\n-from skimage.io import imsave,imread\n-from skimage.segmentation import clear_border, watershed, find_boundaries\n-from scipy.ndimage.filters import uniform_filter\n-from os.path import *\n-from os import listdir, makedirs, remove\n-import pickle\n-import shutil\n-import fnmatch\n-import cv2\n-import sys\n-import argparse\n-import re\n-import copy\n-import datetime\n-from joblib import Parallel, delayed\n-from rowit import WindowView, crop_with_padding_mask\n-from save_tifffile_pyramid import save_pyramid\n-import subprocess\n-import ome_types\n-\n-\n-def imshowpair(A,B):\n- plt.imshow(A,cmap='Purples')\n- plt.imshow(B,cmap='Greens',alpha=0.5)\n- plt.show()\n-\n- \n-def imshow(A):\n- plt.imshow(A)\n- plt.show()\n- \n-def overlayOutline(outline,img):\n- img2 = img.copy()\n- stacked_img = np.stack((img2,)*3, axis=-1)\n- stacked_img[outline > 0] = [65535, 0, 0]\n- imshowpair(img2,stacked_img)\n- \n-def normI(I):\n- Irs=resize(I,(I.shape[0]//10,I.shape[1]//10) );\n- p1 = np.percentile(Irs,10);\n- J = I-p1;\n- p99 = np.percentile(Irs,99.99);\n- J = J/(p99-p1);\n- return J\n-\n-def contour_pm_watershed(\n- contour_pm, sigma=2, h=0, tissue_mask=None,\n- padding_mask=None, min_area=None, max_area=None\n-):\n- if tissue_mask is None:\n- tissue_mask = np.ones_like(contour_pm)\n- padded = None\n- if padding_mask is not None and np.any(padding_mask == 0):\n- contour_pm, padded = crop_with_padding_mask(\n- contour_pm, padding_mask, return_mask=True\n- )\n- tissue_mask = crop_with_padding_mask(\n- tissue_mask, padding_mask\n- )\n- \n- maxima = peak_local_max(\n- extrema.h_maxima(\n- ndi.gaussian_filter(np.invert(contour_pm), sigma=sigma),\n- h=h\n- ),\n- indices=False,\n- footprint=np.ones((3, 3))\n- )\n- maxima = label(maxima).astype(np.int32)\n- \n- # Passing mask into the watershed function will exclude seeds outside\n- # of the mask, which gives fewer and more accurate segments\n- maxima = watershed(\n- contour_pm, maxima, watershed_line=True, mask=tissue_mask\n- ) > 0\n- \n- if min_area is not None and max_area is not None:\n- maxima = label(maxima, connectivity=1).astype(np.int32)\n- areas = np.bincount(maxima.ravel())\n- size_passed = np.arange(areas.size)[\n- np.logical_and(areas > min_area, areas < max_area)\n- ]\n- maxima *= np.isin(maxima, size_passed)\n- np.greater(maxima, 0, out=maxima)\n-\n- if padded is None:\n- return maxima.astype(np.bool)\n- else:\n- padded[padded == 1] = maxima.flatten()\n- return padded.astype(np.bool)\n-\n-def S3AreaSegmenter(nucleiPM, images, TMAmask, threshold,fileprefix,outputPath):\n- nucleiCenters = nucleiPM[:,:,0]\n- TMAmask= (nucleiCenters>np.amax(nucleiCenters)*0.8)*TMAmask\n- area = []\n- area.append(np.sum(np.sum(TMAmask)))\n- for iChan in range(len(images)):\n- image_gauss = gaussian(resize(images[iChan,:,:],(int(0.25*images.shape[1]),int(0.25*images.shape[2]))),1)\n- if threshold ==-1:\n- threshold = threshold_otsu(image_gauss)\n- mask=resize(image_gauss>threshold,(images.shape[1],images.shape[2]),order = 0)*TMAmask\n- area.append(np.sum(np.sum(mask)))\n- os.mk \n- np.savetxt(outputPath + os.path.sep + fileprefix + '_area.csv',(np.transpose(np.asarray(area))),fmt='%10.5f') \n- return TMAmask\n-\n-def getMetadata(path,comm"..b"- cytoFull= tifffile.imread(imagePath, key=iChan)\n- cyto[count,:,:] = cytoFull[int(PMrect[0]):int(PMrect[0]+PMrect[2]), int(PMrect[1]):int(PMrect[1]+PMrect[3])]\n- count+=1 \n- else:\n- cyto=np.empty((len(CytoMaskChan),rect[3],rect[2]),dtype=np.int16)\n- for iChan in CytoMaskChan:\n- cytoFull= tifffile.imread(imagePath, key=iChan)\n- cyto[count,:,:] = cytoFull[int(PMrect[0]):int(PMrect[0]+PMrect[2]), int(PMrect[1]):int(PMrect[1]+PMrect[3])]\n- count+=1\n- cyto = np.amax(cyto,axis=0)\n- cytoplasmMask,nucleiMaskTemp,cellMask = S3CytoplasmSegmentation(nucleiMask,cyto,TMAmask,args.cytoMethod,args.cytoDilation)\n- exportMasks(nucleiMaskTemp,nucleiCrop,outputPath,filePrefix,'nuclei',commit,metadata,args.saveFig,args.saveMask)\n- exportMasks(cytoplasmMask,cyto,outputPath,filePrefix,'cyto',commit,metadata,args.saveFig,args.saveMask)\n- exportMasks(cellMask,cyto,outputPath,filePrefix,'cell',commit,metadata,args.saveFig,args.saveMask)\n- \n- cytoplasmMask,nucleiMaskTemp,cellMask = S3CytoplasmSegmentation(nucleiMask,cyto,TMAmask,'ring',args.cytoDilation)\n- exportMasks(nucleiMaskTemp,nucleiCrop,outputPath,filePrefix,'nucleiRing',commit,metadata,args.saveFig,args.saveMask)\n- exportMasks(cytoplasmMask,cyto,outputPath,filePrefix,'cytoRing',commit,metadata,args.saveFig,args.saveMask)\n- exportMasks(cellMask,cyto,outputPath,filePrefix,'cellRing',commit,metadata,args.saveFig,args.saveMask)\n- \n- elif args.segmentCytoplasm == 'ignoreCytoplasm':\n- exportMasks(nucleiMask,nucleiCrop,outputPath,filePrefix,'nuclei',commit,metadata)\n- cellMask = nucleiMask\n- exportMasks(nucleiMask,nucleiCrop,outputPath,filePrefix,'cell',commit,metadata)\n- cytoplasmMask = nucleiMask\n- \n- detectPuncta = args.detectPuncta\n- if (np.min(detectPuncta)>0):\n- detectPuncta[:] = [number - 1 for number in detectPuncta] #convert 1-based indexing to 0-based indexing \n- if len(detectPuncta) != len(args.punctaSigma):\n- args.punctaSigma = args.punctaSigma[0] * np.ones(len(detectPuncta))\n- \n- \n- if len(detectPuncta) != len(args.punctaSD):\n- args.punctaSD = args.punctaSD[0] * np.ones(len(detectPuncta))\n- \n- counter=0\n- for iPunctaChan in detectPuncta:\n- punctaChan = tifffile.imread(imagePath,key = iPunctaChan)\n- punctaChan = punctaChan[int(PMrect[0]):int(PMrect[0]+PMrect[2]), int(PMrect[1]):int(PMrect[1]+PMrect[3])]\n- spots=S3punctaDetection(punctaChan,cellMask,args.punctaSigma[counter],args.punctaSD[counter])\n- cellspotmask = nucleiMask\n- P = regionprops(cellspotmask,intensity_image = spots ,cache=False)\n- numSpots = []\n- for prop in P:\n- numSpots.append(np.uint16(np.round(prop.mean_intensity * prop.area)))\n- np.savetxt(outputPath + os.path.sep + 'numSpots_chan' + str(iPunctaChan+1) +'.csv',(np.transpose(np.asarray(numSpots))),fmt='%10.5f') \n- edges = 1-(cellMask>0)\n- stacked_img=np.stack((np.uint16((spots+edges)>0)*np.amax(punctaChan),punctaChan),axis=0)\n- \n- \n- outputPathPuncta = outputPath + os.path.sep + filePrefix + os.path.sep + 'punctaChan'+str(iPunctaChan+1) + 'Outlines.ome.tif'\n- \n- # metadata_args = dict(\n- # pixel_sizes=(metadata.physical_size_y, metadata.physical_size_x),\n- # pixel_size_units=('\xc2\xb5m', '\xc2\xb5m'),\n- # software= 's3segmenter v' + commit\n- # )\n- save_pyramid(\n- stacked_img,\n- outputPathPuncta,\n- channel_names=['puncta outlines', 'image channel'],\n- is_mask=False,\n- **metadata\n- ) \n- \n- counter=counter+1 \n-\n" |
| b |
| diff -r 41e8efe8df43 -r 96d0d969ebc9 macros.xml --- a/macros.xml Fri Mar 11 23:37:49 2022 +0000 +++ b/macros.xml Fri Sep 16 20:05:54 2022 +0000 |
| [ |
| @@ -2,25 +2,36 @@ <macros> <xml name="requirements"> <requirements> - <container type="docker">labsyspharm/s3segmenter:@VERSION@</container> + <!-- <requirement type="package" version="3.7">python</requirement> <requirement type="package">scikit-learn</requirement> <requirement version="0.14.2" type="package">scikit-image</requirement> <requirement type="package">matplotlib</requirement> <requirement version="2021.6.6" type="package">tifffile</requirement> <requirement type="package">opencv</requirement> - <!-- <requirement type="package">ome_types</requirement> --> + --> + <container type="docker">labsyspharm/s3segmenter:@TOOL_VERSION@</container> </requirements> </xml> - <xml name="version_cmd"> - <version_command>echo @VERSION@</version_command> - </xml> <xml name="citations"> <citations> </citations> </xml> - <token name="@VERSION@">1.3.12</token> - <token name="@CMD_BEGIN@">python3 $__tool_directory__/S3segmenter.py</token> + <token name="@TOOL_VERSION@">1.3.12</token> + <token name="@VERSION_SUFFIX@">0</token> + <token name="@PROFILE@">19.01</token> + <token name="@CMD_BEGIN@"><![CDATA[ + S3SEG_CMD="" && + if [ -f "/app/S3segmenter.py" ]; then + export S3SEG_CMD="python /app/S3segmenter.py"; + export PYTHONPATH="/app"; + else + export S3SEG_CMD="S3segmenter.py"; + fi && + ]]></token> + <xml name="version_cmd"> + <version_command>@CMD_BEGIN@ $S3SEG_CMD --help</version_command> + </xml> </macros> |
| b |
| diff -r 41e8efe8df43 -r 96d0d969ebc9 rowit.py --- a/rowit.py Fri Mar 11 23:37:49 2022 +0000 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 |
| [ |
| @@ -1,76 +0,0 @@ -import numpy as np -from skimage.util import view_as_windows, montage - - -class WindowView(object): - - def __init__( - self, img_shape, block_size, overlap_size - ): - self.img_shape = img_shape - self.block_size = block_size - self.overlap_size = overlap_size - - self.step_size = block_size - overlap_size - - def window_view_list(self, img, pad_mode='constant'): - half = int(self.overlap_size / 2) - img = np.pad(img, ( - (half, self.padded_shape[0] - self.img_shape[0] - half), - (half, self.padded_shape[1] - self.img_shape[1] - half), - ), mode=pad_mode) - - return self._window_view_list(img) - - def padding_mask(self): - half = int(self.overlap_size / 2) - padding_mask = np.ones(self.img_shape, dtype=np.bool) - padding_mask = np.pad(padding_mask, ( - (half, self.padded_shape[0] - self.img_shape[0] - half), - (half, self.padded_shape[1] - self.img_shape[1] - half), - ), mode='constant', constant_values=0) - return self._window_view_list(padding_mask) - - def reconstruct(self, img_window_view_list): - grid_shape = self.window_view_shape[:2] - - start = int(self.overlap_size / 2) - end = int(self.block_size - start) - - img_window_view_list = img_window_view_list[..., start:end, start:end] - - return montage( - img_window_view_list, grid_shape=grid_shape - )[:self.img_shape[0], :self.img_shape[1]] - - @property - def padded_shape(self): - padded_shape = np.array(self.img_shape) + self.overlap_size - n = np.ceil((padded_shape - self.block_size) / self.step_size) - padded_shape = (self.block_size + (n * self.step_size)).astype(np.int) - return tuple(padded_shape) - - @property - def window_view_shape(self): - return view_as_windows( - np.empty(self.padded_shape), - self.block_size, self.step_size - ).shape - - def _window_view_list(self, img): - return ( - view_as_windows(img, self.block_size, self.step_size) - .reshape(-1, self.block_size, self.block_size) - ) - -def crop_with_padding_mask(img, padding_mask, return_mask=False): - if np.all(padding_mask == 1): - return (img, padding_mask) if return_mask else img - (r_s, r_e), (c_s, c_e) = [ - (i.min(), i.max() + 1) - for i in np.where(padding_mask == 1) - ] - padded = np.zeros_like(img) - img = img[r_s:r_e, c_s:c_e] - padded[r_s:r_e, c_s:c_e] = 1 - return (img, padded) if return_mask else img \ No newline at end of file |
| b |
| diff -r 41e8efe8df43 -r 96d0d969ebc9 s3segmenter.xml --- a/s3segmenter.xml Fri Mar 11 23:37:49 2022 +0000 +++ b/s3segmenter.xml Fri Sep 16 20:05:54 2022 +0000 |
| [ |
| @@ -1,49 +1,49 @@ -<tool id="s3segmenter" name="s3segmenter" version="@VERSION@.0" profile="17.09"> - <description>S3segmenter is a Python-based set of functions that generates single cell (nuclei and cytoplasm) label masks.</description> +<tool id="s3segmenter" name="s3segmenter" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@"> + <description>single cell (nuclei and cytoplasm) label masks.</description> <macros> <import>macros.xml</import> </macros> <expand macro="requirements"/> - @VERSION_CMD@ + <expand macro="version_cmd"/> <command detect_errors="exit_code"><![CDATA[ - ln -s '${imagePath}' ./image.tif; + ln -s '${imagePath}' './image.tif' && #if $contoursClassProbPath - ln -s ${contoursClassProbPath} ./ContoursPM.tif; + ln -s '${contoursClassProbPath}' './ContoursPM.tif' && #end if #if $nucleiClassProbPath - ln -s ${nucleiClassProbPath} ./NucleiPM.tif; + ln -s '${nucleiClassProbPath}' './NucleiPM.tif' && #end if #if $stackProbPath - ln -s ${stackProbPath} ./Probabilities.tif; + ln -s '${stackProbPath}' './Probabilities.tif' && #end if - @CMD_BEGIN@ - --imagePath ./image.tif + @CMD_BEGIN@ \$S3SEG_CMD + --imagePath './image.tif' #if $contoursClassProbPath - --contoursClassProbPath ./ContoursPM.tif + --contoursClassProbPath './ContoursPM.tif' #end if #if $nucleiClassProbPath - --nucleiClassProbPath ./NucleiPM.tif + --nucleiClassProbPath './NucleiPM.tif' #end if #if $stackProbPath - --stackProbPath ./Probabilities.tif + --stackProbPath './Probabilities.tif' #end if --probMapChan $probMapChan --crop $crop_select.crop #if $crop_select.crop == "dearray" - --maskPath $crop_select.maskPath + --maskPath '$crop_select.maskPath' #end if --cytoMethod $cytoMethod @@ -63,16 +63,9 @@ --detectPuncta $adv.detectPuncta --punctaSigma $adv.punctaSigma --punctaSD $adv.punctaSD - - #if not $saveMask - --saveMask - #end if - - #if not $saveFig - --saveFig - #end if - - --outputPath . + $saveMask + $saveFig + --outputPath '.' ]]></command> @@ -94,6 +87,9 @@ <when value="dearray"> <param name="maskPath" type="data" format="tiff" label="TMA Mask File"/> </when> + <when value="noCrop" /> + <when value="autoCrop" /> + <when value="plate" /> </conditional> <param name="cytoMethod" type="select" label="Cyto Method"> @@ -121,17 +117,24 @@ <option value="pixellevel">pixellevel</option> </param> <when value="pixellevel"> - <param name="pixelThreshold" type="float" value="-1.0" Label="Pixel Threshold"/> - <param name="pixelMaskChan" type="text" value="2" Label="Pixel Mask Channel"/> + <param name="pixelThreshold" type="float" value="-1.0" label="Pixel Threshold"/> + <param name="pixelMaskChan" type="text" value="2" label="Pixel Mask Channel"/> </when> + <when value="watershedContourDist"/> + <when value="watershedContourInt"/> + <when value="watershedBWDist"/> + <when value="dilation"/> + <when value="localThreshold"/> + <when value="localMax"/> + <when value="bypass"/> </conditional> <param name="segmentCytoplasm" type="select" label="Segment Cytoplasm"> <option value="segmentCytoplasm">segmentCytoplasm</option> <option selected="true" value="ignoreCytoplasm">ignoreCytoplasm</option> </param> - <param name="saveMask" type="boolean" checked="true" label="Save Mask"/> - <param name="saveFig" type="boolean" checked="true" label="Save Figure"/> + <param argument="saveMask" type="boolean" checked="true" truevalue="--saveMask" falsevalue="" label="Save Mask"/> + <param argument="saveFig" type="boolean" checked="true" truevalue="--saveFig" falsevalue="" label="Save Figure"/> <section name="adv" title="Advanced Options" expanded="false"> <param name="cytoDilation" type="integer" value="5" label="Cyto Dilation"/> @@ -161,7 +164,29 @@ <filter>saveFig is True</filter> </data> </outputs> + <tests> + <test> + <param name="imagePath" value="test.ome.tiff" /> + <param name="stackProbPath" value="stack_probabilities.tiff" /> + <param name="punctaSD" value="4" /> + <output name="cell_mask" ftype="tiff"> + <assert_contents> + <has_size value="6600000" delta="1000000" /> + </assert_contents> + </output> + <output name="nuclei_mask" ftype="tiff"> + <assert_contents> + <has_size value="6600000" delta="1000000" /> + </assert_contents> + </output> + </test> + </tests> <help><![CDATA[ +------------------- +S3segmenter +------------------- +**S3segmenter** is a Python-based set of functions that generates single cell (nuclei and cytoplasm) label masks. + Inputs are: 1. an .ome.tif (preferably flat field corrected) @@ -172,7 +197,6 @@ To segment cytoplasm, the nuclei are in turn used for a marker-controlled watershed segmentation constrained by a cytoplasmic marker such as B-catenin. The channel number of this marker must be specified. A 3-pixel annulus around each nucleus will also be used to segment cytoplasm. The source repository can be found here: https://github.com/HMS-IDAC/S3segmenter -OHSU Wrapper Repo: https://github.com/ohsu-comp-bio/S3segmenter ]]></help> <expand macro="citations" /> </tool> |
| b |
| diff -r 41e8efe8df43 -r 96d0d969ebc9 save_tifffile_pyramid.py --- a/save_tifffile_pyramid.py Fri Mar 11 23:37:49 2022 +0000 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 |
| [ |
| @@ -1,114 +0,0 @@ -import numpy as np -import tifffile -import skimage.transform - -PHYSICAL_SIZE_UNIT = ['Ym', 'Zm', 'Em', 'Pm', 'Tm', 'Gm', 'Mm', 'km', 'hm', 'dam', 'm', 'dm', 'cm', 'mm', 'µm', 'nm', 'pm', 'fm', 'am', 'zm', 'ym', 'Å', 'thou', 'li', 'in', 'ft', 'yd', 'mi', 'ua', 'ly', 'pc', 'pt', 'pixel', 'reference frame'] - -def normalize_image_shape(img): - assert img.ndim in (2, 3), ( - 'image must be 2D (Y, X) or 3D (C, Y, X)' - ) - if img.ndim == 2: - img = img.reshape(1, *img.shape) - assert np.argmin(img.shape) == 0, ( - '3D image must be in (C, Y, X) order' - ) - return img - -def save_pyramid( - out_img, output_path, - pixel_sizes=(1, 1), - pixel_size_units=('µm', 'µm'), - channel_names=None, - software=None, - is_mask=False -): - assert '.ome.tif' in str(output_path) - assert len(pixel_sizes) == len(pixel_size_units) == 2 - assert out_img.ndim in (2, 3), ( - 'image must be either 2D (Y, X) or 3D (C, Y, X)' - ) - - img_shape_ori = out_img.shape - out_img = normalize_image_shape(out_img) - img_shape = out_img.shape - - size_x, size_y = np.array(pixel_sizes, dtype=float) - unit_x, unit_y = pixel_size_units - - assert (unit_x in PHYSICAL_SIZE_UNIT) & (unit_y in PHYSICAL_SIZE_UNIT), ( - f'pixel_size_units must be a tuple of the followings: ' - f'{", ".join(PHYSICAL_SIZE_UNIT)}' - ) - - n_channels = img_shape[0] - if channel_names == None: - channel_names = [f'Channel {i}' for i in range(n_channels)] - else: - if type(channel_names) == str: - channel_names = [channel_names] - n_channel_names = len(channel_names) - assert n_channel_names == n_channels, ( - f'number of channel_names ({n_channel_names}) must match ' - f'number of channels ({n_channels})' - ) - - if software == None: - software = '' - - metadata = { - 'Creator': software, - 'Pixels': { - 'PhysicalSizeX': size_x, - 'PhysicalSizeXUnit': unit_x, - 'PhysicalSizeY': size_y, - 'PhysicalSizeYUnit': unit_y, - }, - 'Channel': {'Name': channel_names}, - - } - - max_size = np.max(img_shape) - subifds = np.ceil(np.log2(max_size / 1024)).astype(int) - - # use optimal tile size for disk space - tile_size = 16*np.ceil( - np.array(img_shape[1:]) / (2**subifds) / 16 - ).astype(int) - options = { - 'tile': tuple(tile_size) - } - - with tifffile.TiffWriter(output_path, bigtiff=True) as tiff_out: - tiff_out.write( - data=out_img, - metadata=metadata, - software=software, - subifds=subifds, - **options - ) - for i in range(subifds): - if i == 0: - down_2x_img = downsize_img_channels(out_img, is_mask=is_mask) - else: - down_2x_img = downsize_img_channels(down_2x_img, is_mask=is_mask) - tiff_out.write( - data=down_2x_img, - subfiletype=1, - **options - ) - - out_img = out_img.reshape(img_shape_ori) - return - -def downsize_channel(img, is_mask): - if is_mask: - return img[::2, ::2] - else: - return skimage.transform.downscale_local_mean(img, (2, 2)).astype(img.dtype) - -def downsize_img_channels(img, is_mask): - return np.array([ - downsize_channel(c, is_mask=is_mask) - for c in img - ]) |
| b |
| diff -r 41e8efe8df43 -r 96d0d969ebc9 test-data/stack_probabilities.tiff |
| b |
| Binary file test-data/stack_probabilities.tiff has changed |
| b |
| diff -r 41e8efe8df43 -r 96d0d969ebc9 test-data/test.ome.tiff |
| b |
| Binary file test-data/test.ome.tiff has changed |