def test_single_kmer(self):
# given
b = get_cortex_builder()
b.with_kmer('AAA 1 ......G.')
b.with_kmer('AAG 1 a.......')
cdb = b.build()
seed = 'AAA'
# when / then
assert [('AAA', 'AAG')] == list(cdb.out_edges(seed))
assert [] == list(cdb.in_edges(seed))
def test_nodes_returns_two_nodes(self):
# given
b = get_cortex_builder()
b.with_kmer('AAA 1 .....C..')
b.with_kmer('AAC 1 a.......')
# when
cortex_graph = b.build()
# then
assert {'AAA', 'AAC'} == set(cortex_graph.nodes())
def test_revcomps_a_kmer():
# given
b = get_cortex_builder()
b.with_kmer('AAA 1 ........')
cdb = b.build()
# when
expect = KmerGraphExpectation(
Interactor(cdb).make_graph_nodes_consistent({'TTT'}).graph)
# then
expect.has_node('TTT')
expect.has_n_nodes(1)
def test_with_2_nodes_has_len_2(self):
# given
b = get_cortex_builder()
b.with_kmer('AAA 1 .....C..')
b.with_kmer('AAC 1 a.......')
# when
cdb = b.build()
ucdb = make_multi_graph(cdb)
# then
assert 2 == len(cdb)
assert 2 == len(ucdb)
def test_remove_nodes(self, removal_nodes):
# given
b = get_cortex_builder()
b.with_kmer('AAA 1 .....C..')
b.with_kmer('AAC 1 a....C..')
b.with_kmer('ACC 1 a.......')
nodes = {'AAC', 'AAA', 'ACC'}
cdb = b.build()
# when
cdb.remove_nodes_from(removal_nodes)
# then
assert (nodes - set(removal_nodes)) == set(cdb.nodes())
def test_in_y_graph_finds_two_paths_of_revcomp(self):
# given
b = get_cortex_builder()
b.with_kmer('CGC 1 .......T')
b.with_kmer('AGC 1 a....CG.')
b.with_kmer('AAG 1 .....C..')
b.with_kmer('GCC 1 a.......')
cdb = b.build()
cdb = Interactor(cdb).make_graph_nodes_consistent(['AAG']).graph
# when
paths = list(Interactor(cdb).all_simple_paths())
# then
assert ['AAGCC', 'AAGCG'] == sorted([str(p.seq) for p in paths])
def test_single_kmer_revcomp_seed(self, seed):
# given
b = get_cortex_builder()
b.with_kmer('AAA 1 ......G.')
b.with_kmer('AAG 1 a.......')
cdb = b.build()
# when
graph = Interactor(cdb).make_graph_nodes_consistent([seed]).graph
# then
if seed == 'AAA':
assert [] == list(graph.in_edges(seed))
assert [('AAA', 'AAG')] == list(graph.out_edges(seed))
else:
assert [('CTT', 'TTT')] == list(graph.in_edges(seed))
assert [] == list(graph.out_edges(seed))
def test_neighbors_of_three_colors_returns_colors_0_and_2(self):
# given
b = get_cortex_builder()
b.with_kmer('AAA 1 1 1 .....C.. ........ .....C..')
b.with_kmer('AAC 1 1 1 a....... ........ a.......')
# when
cdb = b.build()
# then
assert {0, 2} == set(cdb.succ['AAA']['AAC'])
assert set() == set(cdb.pred['AAA']['AAC'])
assert {0, 2} == set(cdb['AAA']['AAC'])
assert set() == set(cdb.succ['AAC']['AAA'])
assert {0, 2} == set(cdb.pred['AAC']['AAA'])
assert set() == set(cdb['AAC']['AAA'])
def test_revcomps_path():
# given
b = get_cortex_builder()
b.with_kmer('CGC 1 .......T')
b.with_kmer('AGC 1 ......G.')
cdb = b.build()
for seed, expected_nodes in [('CGC', ['CGC', 'GCT']),
('GCT', ['CGC', 'GCT']),
('AGC', ['AGC', 'GCG']),
('GCG', ['AGC', 'GCG'])]:
# when
expect = KmerGraphExpectation(
Interactor(cdb).make_graph_nodes_consistent([seed]).graph)
# then
expect.has_nodes(*expected_nodes)
expect.has_n_nodes(2)
def test_remove_node(self, remove_first):
# given
b = get_cortex_builder()
b.with_kmer('AAA 1 .....C..')
b.with_kmer('AAC 1 a.......')
cdb = b.build()
# when
if remove_first:
cdb.remove_node('AAC')
# then
assert {'AAA'} == set(cdb.nodes())
else:
cdb.remove_node('AAA')
# then
assert {'AAC'} == set(cdb.nodes())
def test_keys_y_graph():
# given
b = get_cortex_builder()
b.with_kmer('CGC 1 .......T')
b.with_kmer('AGC 1 a....CG.')
b.with_kmer('AAG 1 .....C..')
b.with_kmer('GCC 1 a.......')
expected_nodes1 = ['CGC', 'GCT', 'CTT', 'GGC']
expected_nodes2 = ['AAG', 'AGC', 'GCG', 'GCC']
for expected_nodes in [expected_nodes1, expected_nodes2]:
for seed in expected_nodes:
cdb = b.build()
# when
expect = KmerGraphExpectation(
Interactor(cdb).make_graph_nodes_consistent([seed]).graph)
# then
expect.has_nodes(*expected_nodes)
def test_revcomps_many_kmers(data, num_kmers, kmer_size):
# given
kmers = {}
for _ in range(num_kmers):
kmer_string = data.draw(kmer_strings(min_size=kmer_size, max_size=kmer_size))
kmers[lexlo(kmer_string)] = kmer_string
b = get_cortex_builder()
for kmer in kmers.keys():
b.with_kmer('{} 1 ........'.format(kmer))
cdb = b.build()
# when
expect = KmerGraphExpectation(
Interactor(cdb).make_graph_nodes_consistent(set(kmers.values())).graph)
# then
for kmer_string in kmers.values():
expect.has_node(kmer_string)
expect.has_n_nodes(len(kmers))
def test_gets_correct_neighbors_of_kmer(self):
# given
b = get_cortex_builder()
b.with_kmer('AAC 1 .......T')
b.with_kmer('ACT 1 a.....G.')
b.with_kmer('CAG 1 .......T')
cdb = b.build()
seed = 'AAC'
# when
graph = Interactor(cdb).make_graph_nodes_consistent([seed]).graph
# then
assert ['CTG'] == list(graph['ACT'])
assert ['CTG'] == list(graph.succ['ACT'])
assert ['AAC'] == list(graph.pred['ACT'])
assert [('ACT', 'CTG')] == list(graph.out_edges('ACT'))
assert [('AAC', 'ACT')] == list(graph.in_edges('ACT'))
assert [] == list(graph['CTG'])
assert [] == list(graph.succ['CTG'])
assert ['ACT'] == list(graph.pred['CTG'])
assert [] == list(graph.out_edges('CTG'))
assert [('ACT', 'CTG')] == list(graph.in_edges('CTG'))