def intra_vs_inter_gene_dist_hist(df):
"""This function compares the intra vs inter gene distance score
by dropping features:
'original': without dropping any features
'cells-with': The proportion of cells with each nucleus counts(each nucleus
counts means that several nucleus count groups can be formed because a
number of cells have the same number of nuclei)
'frac': fraction of nuclei that expresses a particular transcription factor
'd-neighbor': distances between the n nearest neighbors in an image of labeled
nuclei
'theta': angles between nearest neighbors in an image of labeled nuclei
'cos': cosine of angles between nearest neighbors in an image of labeled nuclei
'sin': cosine of angles between nearest neighbors in an image of labeled nuclei
'adaptive': If there is large variation in the background intensity, adaptive or
local threshold is used
'intensity': Segment objects based on intensity threshold and compute properties
'extent': the spatial extent of the well defined rigid objects is the complete
object itself
'euler_number': it is a scalar whose value is the total number of objects in the
image minus the total number of holes in those objects.
'eccentricity': how elongated it is
'area': area of an object
'otsu-threshold': otsu's method is used to automatically perform clustering-based
image thresholding or the reduction of a graylevel image to a binary image.
"""
names = [
'original', 'cells-with', 'frac', 'd-neighbor', 'theta', 'cos', 'sin',
'adaptive', 'solidity', 'intensity', 'extent', 'euler_number',
'eccentricity', 'area', 'otsu-threshold'
]
t_stat_p_comp = []
t_stat_comp = []
ks_stat_p_comp = []
ks_stat_comp = []
coll = MongoClient()['myofusion_test']['wells_test']
for pattern in names:
reduced_column = [c for c in df.columns if pattern not in c.lower()]
reduced_df = df[reduced_column]
intra, inter = metric.gene_distance_score(reduced_df, coll)
t_stat, t_stat_p = scipy.stats.ttest_ind(
intra, inter) # t-test stat and Pvalues
t_stat_p_comp.append(t_stat_p)
t_stat_comp.append(t_stat)
ks_stat, ks_stat_p = scipy.stats.ks_2samp(
intra,
inter) # Computes the Kolmogorov-Smirnov statistic on 2 samples.
ks_stat_p_comp.append(ks_stat_p)
ks_stat_comp.append(ks_stat)
final_dataframe = DataFrame({
't_stat_p_comp': t_stat_p_comp,
't_stat_comp': t_stat_comp,
'ks_stat_p_comp': ks_stat_p_comp,
'ks_stat_comp': ks_stat_comp
})
final_dataframe.index = names
return final_dataframe
def test_gene_distance_score2():
intra, inter = metrics.gene_distance_score(test_data, collection)
assert np.mean(intra) < np.mean(inter)
def test_gene_distance_score2():
intra, inter = metrics.gene_distance_score(test_data, collection)
assert np.mean(intra) < np.mean(inter)
def test_gene_distance_score():
expected_intra = []
for i in range(0, 4):
expected_intra.append(test_distance[2*i][2*i+1])
intra, inter = metrics.gene_distance_score(test_data, collection)
np.testing.assert_array_almost_equal(expected_intra, intra, decimal=4)
def test_gene_distance_score():
expected_intra = []
for i in range(0, 4):
expected_intra.append(test_distance[2 * i][2 * i + 1])
intra, inter = metrics.gene_distance_score(test_data, collection)
np.testing.assert_array_almost_equal(expected_intra, intra, decimal=4)
