Don’t repeat yourself

Sometimes, we copy & paste code multiple times to process images quickly. Long-term this code duplication is not good for code quality, because if we want to change something, we need to change it in muliple places and may forget about some. Thus, preventing repetitive code is key. The software design priniciple is called Don’t repeat yourself.

import pyclesperanto_prototype as cle

image = cle.imread("../../data/BBBC007_batch/17P1_POS0013_D_1UL.tif")
labels = cle.voronoi_otsu_labeling(image, spot_sigma=3)
number_of_nuclei = labels.max()
number_of_nuclei

44.0

image = cle.imread("../../data/BBBC007_batch/20P1_POS0005_D_1UL.tif")
labels = cle.voronoi_otsu_labeling(image, spot_sigma=3)
number_of_nuclei = labels.max()
number_of_nuclei

41.0

image = cle.imread("../../data/BBBC007_batch/20P1_POS0007_D_1UL.tif")
labels = cle.voronoi_otsu_labeling(image, spot_sigma=3)
number_of_nuclei = labels.max()
number_of_nuclei

73.0

If we wanted to now see how changing the spot_sigma parameter above influences the result, we would need to change this value three times. When the code becomes even longer, it may happend that we forget to change it in one place.

For-loops instead of code repetition

One way of preventing code-repetition are for-loops.

folder = "../../data/BBBC007_batch/"
files = ["17P1_POS0013_D_1UL.tif",
        "20P1_POS0005_D_1UL.tif",
        "20P1_POS0007_D_1UL.tif"]

for file in files:
    image = cle.imread(folder + file)
    labels = cle.voronoi_otsu_labeling(
                    image, 
                    spot_sigma=3)
    number_of_nuclei = labels.max()
    print(file, number_of_nuclei)

17P1_POS0013_D_1UL.tif 44.0
20P1_POS0005_D_1UL.tif 41.0
20P1_POS0007_D_1UL.tif 73.0

Functions instead of code-repetion

We can gain even more flexibility by introducing so called helper functions, that help us by doing one dedicated thing, for example counting nuclei:

def count_nuclei(image, spot_sigma=3):
    labels = cle.voronoi_otsu_labeling(
                    image, 
                    spot_sigma=spot_sigma)
    number_of_nuclei = labels.max()
    
    return number_of_nuclei

count_nuclei(cle.imread(folder + files[0]))

44.0

count_nuclei(cle.imread(folder + files[1]))

41.0

count_nuclei(cle.imread(folder + files[2]))

73.0

count_nuclei(cle.imread(folder + files[2]), spot_sigma=5)

68.0