def _check_fp():
_graph = graph.shallow_clone()
funcs = check_fp_collector()
sequences = consume_collector()
for sequence in sequences.values():
for kmer in kmers(sequence, ksize):
if _graph.get(kmer):
print(sequences.items)
check_fp_xfail('k-mer from a sequence is false-positive',
request)
for src_fixture, sequence_bundle in sequences.items.items():
for sequence in sequence_bundle:
_graph.add(sequence)
fp_funcs = funcs.items[src_fixture]
for func, result in fp_funcs:
if func(_graph) != result:
print(sequences.items)
check_fp_xfail(
'neighbor is false-positive from {}'.format(
src_fixture), request)
check_fp_pass(list(funcs.items.keys()))
def test_random_sequence(random_sequence, ksize, length):
seq1 = random_sequence()
seq2 = random_sequence(exclude=seq1)
assert len(seq1) == 20
assert len(seq2) == 20
for kmer in kmers(seq1, ksize):
assert kmer not in seq2
def test_snp_bubble(snp_bubble, ksize, length):
(wildtype, snp), L, R = snp_bubble()
assert wildtype[:L + ksize] == snp[:L + ksize]
assert wildtype[R:] == snp[R:]
assert wildtype[R - 1] != snp[R - 1]
for kmer in kmers(wildtype[L + 1:R], ksize):
assert kmer not in snp
def count_decision_nodes(sequence, graph, ksize):
'''Get the degree distribution of nodes with degree more than 2.
'''
dnodes = {}
for i, kmer in enumerate(kmers(sequence, ksize)):
d = (graph.left_degree(kmer), graph.right_degree(kmer))
ld, rd = d
if ld > 1 or rd > 1:
dnodes[d] = dnodes.get(d, 0) + 1
print(i, d)
return dnodes
def get(exclude=None, length=length):
try:
sequence = get_random_sequence(length,
ksize,
exclude=exclude,
seen=global_seen)
except ValueError:
pytest.xfail(
'Failed to generate sequence of given length and k-size')
for kmer in kmers(sequence, ksize - 1):
global_seen.add(kmer)
return sequence
def _circular_key():
loop = linear_path()
loop = loop + loop[:ksize - 1]
if pivot in (length - ksize, length - ksize - 1):
_pivot = pivot + ksize - 1
else:
_pivot = pivot
loop_kmers = list(kmers(loop, ksize))
tail_kmers = [
loop_kmers[i % len(loop_kmers)]
for i in range(_pivot + 1, _pivot + 1 + ksize)
]
tail = ''.join((kmer[0] for kmer in tail_kmers))
tail = mutate_position(tail, -1)
consume_collector(loop, tail)
check_fp_collector((lambda G: count_decision_nodes(loop, G, ksize), {
(1, 2): 1
}))
return (loop, tail), _pivot
def test_kmers(ksize):
S = 'A' * ksize + 'T'
res = list(kmers(S, ksize))
assert res[0] == 'A' * ksize
assert res[-1] == ('A' * (ksize - 1)) + 'T'