def test_setGeneName(self):
"""
OK.
"""
test_gene1 = 'E:/Master/cophenetic_dists/ENSG00000188324___OR6C6___CopD.csv'
test_gene2 = 'E:/Master/cophenetic_dists/ENSG00000000938___FGR___CopD.csv'
pop_info = 'C:/Users/norab/Master/Data/real_tree_data/phydist_population_classes.tsv'
tree1 = treeInfo()
tree1.setup(test_gene1, pop_info)
tree2 = treeInfo()
tree2.setup(test_gene2, pop_info)
assert tree1.getGeneName() == "ENSG00000188324___OR6C6"
assert tree2.getGeneName() == "ENSG00000000938___FGR"
def test_makePsuedoPops(self):
"""
OK
"""
test_gene = 'C:\\Users\\norab\\Master\\Data\\real_tree_data\\dist_mat_subset\\ENSG00000001167___NFYA___CopD.csv'
pop_info = 'C:/Users/norab/Master/Data/real_tree_data/phydist_population_classes.tsv'
num_pops_tests = [6, 17]
for num_pops in num_pops_tests:
tree = treeInfo()
tree.setup(test_gene, pop_info)
info = tree.getSampleInfo()
dist_mat = tree.getDistMat()
tree.makePsuedoPops(num_pops)
num_samples = len(dist_mat) / num_pops
all_groups = []
for ind, el in info.items():
all_groups.append(el[-1])
all_groups = pd.Series(all_groups)
all_groups.value_counts()
assert len(all_groups) == len(dist_mat)
assert sum(all_groups.value_counts()) == len(dist_mat)
num_samples = int(round(num_samples))
for val in all_groups.value_counts():
assert val in [num_samples, num_samples + 1, num_samples - 1]
def test_shuffleSampleInfo(self):
test_gene_small = 'C:\\Users\\norab\\Master\\Data\\real_tree_data\\dist_mat_test\\FGR_10x10.csv'
pop_info = 'C:/Users/norab/Master/Data/real_tree_data/phydist_population_classes.tsv'
tree = treeInfo()
tree.setup(test_gene_small, pop_info)
# dist_mat = tree.getDistMat() -- used for visual inspection
info1 = tree.getSampleInfo()
info1 = info1.deepcopy()
tree.shuffleSampleInfo()
info2 = tree.getSampleInfo()
assert info1 != info2
def test_getSampleInfo(self):
"""
OK.
"""
test_gene_small = 'C:\\Users\\norab\\Master\\Data\\real_tree_data\\dist_mat_test\\FGR_10x10.csv'
pop_info = 'C:/Users/norab/Master/Data/real_tree_data/phydist_population_classes.tsv'
tree = treeInfo()
tree.setup(test_gene_small, pop_info)
# dist_mat = tree.getDistMat() -- used for visual inspection
info = tree.getSampleInfo()
assert info[2][1] == 'EUR'
assert info[2][2] == 'GBR'
assert list(info.items())[-1][-1][-1] == 'ESN'
def test_setSampleInfo(self):
"""
OK.
"""
test_gene = 'E:/Master/cophenetic_dists/ENSG00000188324___OR6C6___CopD.csv'
pop_info = 'C:/Users/norab/Master/Data/real_tree_data/phydist_population_classes.tsv'
tree = treeInfo()
tree.setup(test_gene, pop_info)
# dist_mat = tree.getDistMat() -- used for visual inspection
sample_info = tree.getSampleInfo()
assert isinstance(sample_info, dict)
assert isinstance(sample_info[2], list)
assert sample_info[0] == ['EUR___GBR___HG00261', 'EUR', 'GBR']
assert sample_info[200] == ['EAS___CHS___HG00409', 'EAS', 'CHS']
def test_setPopInfo(self):
"""
OK.
"""
pop_info_file = 'C:/Users/norab/Master/Data/real_tree_data/phydist_population_classes.tsv'
tree = treeInfo()
tree.setPopInfo(pop_info_file)
pop_info = tree.getPopInfo()
sup = pop_info[0]
sub = pop_info[1]
assert sup == {'EUR', 'EAS', 'SAS', 'AFR', 'AMR'}
assert sub == {
'MXL', 'PUR', 'TSI', 'PEL', 'PJL', 'MSL', 'CHB', 'ASW', 'ESN',
'STU', 'IBS', 'BEB', 'ACB', 'YRI', 'ITU', 'GWD', 'CHS', 'CDX',
'GBR', 'KHV', 'GIH', 'FIN', 'LWK', 'JPT', 'CLM', 'CEU'
}