Skip to content

generateCSScore

Short Description

The generateCSScore function calculates CSPOT Score for each cell by using both the generated probability masks and pre-computed segmentation masks as inputs

Function

generateCSScore(probabilityMaskPath, segmentationMaskPath, feature='median', verbose=True, markerNames=None, projectDir=None)

Parameters:

Name Type Description Default
probabilityMaskPath str

Supply the path of the probability map image produced by dlModelPredict.

 required
segmentationMaskPath str

Supply the path of the pre-computed segmentation mask.

 required
feature str

Calculates the mean or median CSPOT Score for each cell.

 'median'
verbose bool

If True, print detailed information about the process to the console.

True
markerNames list

The program searches for marker names in the meta data (description section) of the tiff files created by csPredict by default. If the meta data is lost due to user modifications, provide the marker names for each channel/layer in the probabilityMaskPath here.

 None
projectDir str

Provide the path to the output directory. The result will be located at projectDir/CSPOT/csScore/.

 None

Returns:

Name Type Description
CSV dataframe

The .csv file containing the csScore is stored in the provided projectDir.
Example 
# global path
projectDir = '/Users/aj/Documents/cspotExampleData'

# Path to all the files that are necessary files for running generateCSScore
segmentationPath = projectDir + '/segmentation/exampleSegmentationMask.tif'
probabilityMaskPath = projectDir + '/CSPOT/csPredict/exampleImage_cspotPredict.ome.tif'

cs.generateCSScore(probabilityMaskPath=probabilityMaskPath,
              segmentationMaskPath=segmentationPath,
              feature='median',
              projectDir=projectDir)

# Same function if the user wants to run it via Command Line Interface
python generateCSScore.py             --probabilityMaskPath /Users/aj/Documents/cspotExampleData/CSPOT/csPredict/exampleImage_cspotPredict.ome.tif             --segmentationMaskPath /Users/aj/Documents/cspotExampleData/segmentation/exampleSegmentationMask.tif             --projectDir /Users/aj/Documents/cspotExampleData
Source code in cspot/generateCSScore.py 
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
def generateCSScore (probabilityMaskPath,
                         segmentationMaskPath,
                         feature='median',
                         verbose=True,
                         markerNames=None,
                         projectDir=None):

    """
Parameters:
    probabilityMaskPath (str):
        Supply the path of the probability map image produced by `dlModelPredict`.

    segmentationMaskPath (str):
        Supply the path of the pre-computed segmentation mask.

    feature (str, optional):
        Calculates the `mean` or `median` CSPOT Score for each cell.

    verbose (bool, optional):
        If True, print detailed information about the process to the console.  

    markerNames (list, optional):
        The program searches for marker names in the meta data (description section)
        of the tiff files created by `csPredict` by default. If the meta data
        is lost due to user modifications, provide the marker names for each
        channel/layer in the `probabilityMaskPath` here.

    projectDir (str, optional):
        Provide the path to the output directory. The result will be located at
        `projectDir/CSPOT/csScore/`.

Returns:
    CSV (dataframe):
        The `.csv` file containing the `csScore` is stored in the provided projectDir.

Example:
        ```python

        # global path
        projectDir = '/Users/aj/Documents/cspotExampleData'

        # Path to all the files that are necessary files for running generateCSScore
        segmentationPath = projectDir + '/segmentation/exampleSegmentationMask.tif'
        probabilityMaskPath = projectDir + '/CSPOT/csPredict/exampleImage_cspotPredict.ome.tif'

        cs.generateCSScore(probabilityMaskPath=probabilityMaskPath,
                      segmentationMaskPath=segmentationPath,
                      feature='median',
                      projectDir=projectDir)

        # Same function if the user wants to run it via Command Line Interface
        python generateCSScore.py \
            --probabilityMaskPath /Users/aj/Documents/cspotExampleData/CSPOT/csPredict/exampleImage_cspotPredict.ome.tif \
            --segmentationMaskPath /Users/aj/Documents/cspotExampleData/segmentation/exampleSegmentationMask.tif \
            --projectDir /Users/aj/Documents/cspotExampleData

        ```

    """

    # read the seg mask
    segM = tifffile.imread(pathlib.Path(segmentationMaskPath))
    probM = tifffile.imread(pathlib.Path(probabilityMaskPath))

    #probs = []
    #for i in range(len(probM)):
    #    pospix = len(probM[i][(probM[i] / 255) > 0.5]) / (probM[i].shape[0] * probM[i].shape[1])
    #    probs.append(pospix)

    if len(probM.shape) > 2:
        probM = np.moveaxis(probM, 0, -1)

    def median_intensity(mask, img):
        return np.median(img[mask])

    # quantify
    if verbose is True:
        print("Quantifying the probability masks")
    quantTable = pd.DataFrame(measure.regionprops_table(segM, intensity_image=probM,
                                                        properties=['label','mean_intensity'],
                                                        extra_properties=[median_intensity])).set_index('label')

    # keep only median
    if feature == 'median':
        quantTable = quantTable.filter(regex='median')
    if feature == 'mean':
        quantTable = quantTable.filter(regex='mean')

    # read the channel names from the tiffile
    tiff = tifffile.TiffFile(probabilityMaskPath)
    try:
        root = ET.fromstring(tiff.pages[0].description)
        # parse the ome XML
        namespace = None
        for elem in root.iter():
            if "Channel" in elem.tag:
                namespace = {"ome": elem.tag.split("}")[0][1:]}
                break
        channel_names = [channel.get("Name") for channel in root.findall(".//ome:Channel", namespace)]
        quantTable.columns = channel_names
        #omexml_string = ast.literal_eval(tiff.pages[0].description)
        #channel_names = omexml_string['Channel']['Name']
    except:
        pass
    if markerNames is not None:
        channel_names = markerNames
    else:
        channel_names = list(quantTable.columns)

    # assign channel names
    quantTable.columns = channel_names

    # build a division vector
    # this is to make sure the low probs are not amplified; chosing 154 as it is 0.6P
    #div_val = []
    #for i in quantTable.columns:
    #    div_val.append(255 if quantTable[i].max() < 154 else  quantTable[i].max())

    # conver to prob
    #quantTable = quantTable / div_val
    quantTable = quantTable / 255

    # identify markers that failed
    #sns.distplot(quantTable['ECAD'])


    # if projectDir is given
    if projectDir is None:
        projectDir = os.getcwd()

    # final path to save results
    finalPath = pathlib.Path(projectDir + '/CSPOT/csScore/')
    if not os.path.exists(finalPath):
        os.makedirs(finalPath)

    # file name
    file_name = pathlib.Path(probabilityMaskPath).stem + '.csv'
    quantTable.to_csv(finalPath / file_name)

    # Finish Job
    if verbose is True:
        print('csScore is ready, head over to' + str(projectDir) + '/CSPOT/csScore" to view results')