def kmer_records(draw, kmer_size, num_colors, kmer_strings=dna_sequences):
kmer = draw(kmer_strings(min_size=kmer_size, max_size=kmer_size))
coverage = tuple(
draw(
s.lists(s.integers(min_value=1, max_value=MAX_UINT),
min_size=num_colors,
max_size=num_colors)))
edges = np.array(draw(
s.lists(s.lists(s.integers(min_value=0, max_value=1),
min_size=8,
max_size=8),
min_size=num_colors,
max_size=num_colors)),
dtype=np.uint8)
edges = [EdgeSet(np.concatenate((e[:4], e[::-1][:4]))) for e in edges]
return KmerRecord(kmer, coverage, edges)
def test_empty_kmer(self):
es = EdgeSet(np.zeros(8))
for as_revcomp in [True, False]:
assert es.to_str(as_revcomp=as_revcomp) == '........'
def test_raises_on_non_lexlo_kmer(self):
es = EdgeSet(np.zeros(8))
with pytest.raises(AssertionError):
es.get_outgoing_kmers('TTT')
with pytest.raises(AssertionError):
es.get_incoming_kmers('TTT')
def test_outgoing_strings_does_not_return_lexicographically_lowest_kmer(self):
es = EdgeSet(np.zeros(8))
es.add_edge('G')
assert ['CGG'] == list(es.get_outgoing_kmer_strings('ACG'))
def test_outgoing_returns_lexicographically_lowest_kmers(self):
es = EdgeSet(np.zeros(8))
es.add_edge('G')
assert ['CCG'] == es.get_outgoing_kmers('ACG')
def test_incoming_strings_does_not_return_lexicographically_lowest_kmers(self):
es = EdgeSet(np.zeros(8))
es.add_edge('t')
assert ['TTA'] == list(es.get_incoming_kmer_strings('TAA'))
def test_with_all_true(self):
es = EdgeSet(np.ones(8))
for letter in 'acgtACGT':
assert es.is_edge(letter)
def test_with_none_true(self):
es = EdgeSet(np.zeros(8))
for letter in 'acgtACGT':
assert not es.is_edge(letter)
def test_no_incoming_all_outgoing(self):
es = EdgeSet(np.concatenate([np.zeros(4), np.ones(4)]))
assert 0 == len(es.get_incoming_kmers('AAA'))
assert {'AAA', 'AAC', 'AAG', 'AAT'} == set(es.get_outgoing_kmers('AAA'))
def test_all_incoming_no_outgoing(self):
es = EdgeSet(np.concatenate([np.ones(4), np.zeros(4)]))
assert es.get_incoming_kmers('AAA') == ['AAA', 'CAA', 'GAA', 'TAA']
assert 0 == len(es.get_outgoing_kmers('AAA'))
def test_no_incoming_or_outgoing(self):
es = EdgeSet(np.zeros(8))
assert 0 == len(es.get_incoming_kmers('AAA'))
assert 0 == len(es.get_outgoing_kmers('AAA'))
def test_with_all_incoming_and_no_outgiong(self):
es = EdgeSet(np.concatenate([np.ones(4), np.zeros(4)]))
assert list(es.incoming) == [1, 1, 1, 1]
assert list(es.outgoing) == [0, 0, 0, 0]
def test_works(self):
es = EdgeSet(np.ones(8))
for edge_idx in range(8):
assert es[edge_idx]
def test_removes_each_edge(self):
es = EdgeSet(np.ones(8))
for letter in 'acgtACGT':
assert es.is_edge(letter)
es.remove_edge(letter)
assert not es.is_edge(letter)
def test_adds_each_edge(self):
es = EdgeSet(np.zeros(8))
for letter in 'acgtACGT':
assert not es.is_edge(letter)
es.add_edge(letter)
assert es.is_edge(letter)
def test_incoming_returns_lexicographically_lowest_kmers(self):
es = EdgeSet(np.zeros(8))
es.add_edge('t')
assert ['TAA'] == es.get_incoming_kmers('TAA')
def as_edge_set(edge_set_string):
return EdgeSet(edge_set_string_to_array(edge_set_string))
def test_with_a_and_c(self):
es = EdgeSet(np.zeros(8))
es.add_edge('A')
es.add_edge('c')
assert '.c..A...' == es.to_str()
assert '...T..g.' == es.to_str(as_revcomp=True)
