def test_named_indices(self):
parent_regions = Regions(pd.DataFrame({'chromosome': ['chr1', 'chr10'], 'start': [100, 200], 'end': [160, 220]},
index=['foo', 'bar']))
# Edges
edge_regions = Regions(pd.DataFrame({'chromosome': ['chr1', 'chr10'], 'start': [100, 215], 'end': [110, 220]},
index=['foo', 'bar']))
correct_ans = {'foo': np.array([0, 1]), 'bar': np.array([3])}
self.assert_numpy_dicts_equal(correct_ans, edge_regions.as_bins_of(parent_regions, resolution=5))
# Middle
mid_regions = Regions(pd.DataFrame({'chromosome': ['chr1', 'chr10'], 'start': [140, 208], 'end': [145, 212]},
index=['foo', 'bar']))
correct_ans = {'foo': np.array([8]), 'bar': np.array([1, 2])}
self.assert_numpy_dicts_equal(correct_ans, mid_regions.as_bins_of(parent_regions, resolution=5))
# Middle, without one index
mid_regions2 = Regions(pd.DataFrame({'chromosome': ['chr10'], 'start': [208], 'end': [212]}, index=['bar']))
correct_ans = {'bar': np.array([1, 2])}
self.assert_numpy_dicts_equal(correct_ans, mid_regions2.as_bins_of(parent_regions, resolution=5))
length_one_regions = Regions(pd.DataFrame({'chromosome': ['chr1'], 'start': [140], 'end': [141]}, index=['foo']))
correct_ans = {'foo': np.array([8])}
self.assert_numpy_dicts_equal(correct_ans, length_one_regions.as_bins_of(parent_regions, resolution=5))
regions_not_in_parent = Regions(pd.DataFrame({'chromosome': ['chr1', 'chr2'], 'start': [140, 200], 'end': [141, 300]},
index=['foo', 'baz']))
correct_ans = {'foo': np.array([8])}
self.assert_numpy_dicts_equal(correct_ans, regions_not_in_parent.as_bins_of(parent_regions, resolution=5))
def test_bins_calculated_correctly(self):
parent_regions = Regions(
pd.DataFrame({
'chromosome': ['chr1', 'chr10'],
'start': [100, 200],
'end': [160, 220]
}))
# Edges
edge_regions = Regions(
pd.DataFrame({
'chromosome': ['chr1', 'chr10'],
'start': [100, 215],
'end': [110, 220]
}))
correct_ans = {0: np.array([0, 1]), 1: np.array([3])}
self.assert_numpy_dicts_equal(
correct_ans, edge_regions.as_bins_of(parent_regions, resolution=5))
# Middle
mid_regions = Regions(
pd.DataFrame({
'chromosome': ['chr1', 'chr10'],
'start': [140, 208],
'end': [145, 212]
}))
correct_ans = {0: np.array([8]), 1: np.array([1, 2])}
self.assert_numpy_dicts_equal(
correct_ans, mid_regions.as_bins_of(parent_regions, resolution=5))
# Middle, without one index
mid_regions2 = Regions(
pd.DataFrame(
{
'chromosome': ['chr10'],
'start': [208],
'end': [212]
},
index=[1]))
correct_ans = {1: np.array([1, 2])}
self.assert_numpy_dicts_equal(
correct_ans, mid_regions2.as_bins_of(parent_regions, resolution=5))
length_one_regions = Regions(
pd.DataFrame({
'chromosome': ['chr1'],
'start': [140],
'end': [141]
}))
correct_ans = {0: np.array([8])}
self.assert_numpy_dicts_equal(
correct_ans,
length_one_regions.as_bins_of(parent_regions, resolution=5))
def test_bins_calculated_correctly(self):
parent_regions = Regions(pd.DataFrame({'chromosome': ['chr1', 'chr10'], 'start': [100, 200], 'end': [160, 220]}))
# Edges
edge_regions = Regions(pd.DataFrame({'chromosome': ['chr1', 'chr10'], 'start': [100, 215], 'end': [110, 220]}))
correct_ans = {0: np.array([0, 1]), 1: np.array([3])}
self.assert_numpy_dicts_equal(correct_ans, edge_regions.as_bins_of(parent_regions, resolution=5))
# Middle
mid_regions = Regions(pd.DataFrame({'chromosome': ['chr1', 'chr10'], 'start': [140, 208], 'end': [145, 212]}))
correct_ans = {0: np.array([8]), 1: np.array([1, 2])}
self.assert_numpy_dicts_equal(correct_ans, mid_regions.as_bins_of(parent_regions, resolution=5))
# Middle, without one index
mid_regions2 = Regions(pd.DataFrame({'chromosome': ['chr10'], 'start': [208], 'end': [212]}, index=[1]))
correct_ans = {1: np.array([1, 2])}
self.assert_numpy_dicts_equal(correct_ans, mid_regions2.as_bins_of(parent_regions, resolution=5))
length_one_regions = Regions(pd.DataFrame({'chromosome' : ['chr1'], 'start': [140], 'end': [141]}))
correct_ans = {0: np.array([8])}
self.assert_numpy_dicts_equal(correct_ans, length_one_regions.as_bins_of(parent_regions, resolution=5))
def test_named_indices(self):
parent_regions = Regions(
pd.DataFrame(
{
'chromosome': ['chr1', 'chr10'],
'start': [100, 200],
'end': [160, 220]
},
index=['foo', 'bar']))
# Edges
edge_regions = Regions(
pd.DataFrame(
{
'chromosome': ['chr1', 'chr10'],
'start': [100, 215],
'end': [110, 220]
},
index=['foo', 'bar']))
correct_ans = {'foo': np.array([0, 1]), 'bar': np.array([3])}
self.assert_numpy_dicts_equal(
correct_ans, edge_regions.as_bins_of(parent_regions, resolution=5))
# Middle
mid_regions = Regions(
pd.DataFrame(
{
'chromosome': ['chr1', 'chr10'],
'start': [140, 208],
'end': [145, 212]
},
index=['foo', 'bar']))
correct_ans = {'foo': np.array([8]), 'bar': np.array([1, 2])}
self.assert_numpy_dicts_equal(
correct_ans, mid_regions.as_bins_of(parent_regions, resolution=5))
# Middle, without one index
mid_regions2 = Regions(
pd.DataFrame(
{
'chromosome': ['chr10'],
'start': [208],
'end': [212]
},
index=['bar']))
correct_ans = {'bar': np.array([1, 2])}
self.assert_numpy_dicts_equal(
correct_ans, mid_regions2.as_bins_of(parent_regions, resolution=5))
length_one_regions = Regions(
pd.DataFrame({
'chromosome': ['chr1'],
'start': [140],
'end': [141]
},
index=['foo']))
correct_ans = {'foo': np.array([8])}
self.assert_numpy_dicts_equal(
correct_ans,
length_one_regions.as_bins_of(parent_regions, resolution=5))
regions_not_in_parent = Regions(
pd.DataFrame(
{
'chromosome': ['chr1', 'chr2'],
'start': [140, 200],
'end': [141, 300]
},
index=['foo', 'baz']))
correct_ans = {'foo': np.array([8])}
self.assert_numpy_dicts_equal(
correct_ans,
regions_not_in_parent.as_bins_of(parent_regions, resolution=5))