def test_modify_query_ends(query_subject_example):
query, subject = query_subject_example
c = Contig(query, subject, "Example")
le, re = query.left_end + 15, query.right_end - 10
c.modify_query_ends(le, re)
assert c.query.left_end == le
assert c.query.right_end == re
assert c.query.length == c.subject.length
def test_contig_copy(query_subject_example):
query, subject = query_subject_example
c = Contig(query, subject, "Example")
c_copy = c.copy()
assert c_copy is not c
c2_copy = copy(c)
assert c2_copy is not c
assert c2_copy.contig_id != c.contig_id
assert c_copy.contig_id != c.contig_id
assert c_copy.contig_id != c2_copy.contig_id
def test_modify_subject(query_subject_example):
query, subject = query_subject_example
c = Contig(query, subject, "Example")
qs, qe = query.start-10, query.end-10
ss, se = subject.start+10, subject.end+10
c.modify_subject(ss, se)
assert c.query.start == c.query.context.translate_pos(qs)
assert c.query.end == c.query.context.translate_pos(qe)
assert c.query.length == c.subject.length
assert c.subject.start == c.subject.context.translate_pos(ss)
assert c.subject.end == c.subject.context.translate_pos(se)
def test_subquery():
c1 = Context(10000, True)
c2 = Context(5000, True)
q1 = ContigRegion(100, 300, context=c1, forward=True)
s1 = ContigRegion(500, 700, context=c2, forward=True)
contig = Contig(q1, s1, "BLAST")
sub_query1 = contig.sub_query(200, 250)
sub_query2 = contig.sub_query(220, 300)
assert len(sub_query1) == 51
assert len(sub_query2) == 81
def test_contig_reverse_direction(query_subject_example):
"""This contig should reverse the direction of contig"""
query, subject = query_subject_example
query.sequence = "ACGTAGTGCTGTSCGGCTGTGATGCGT"
subject.sequence = "ACGTAGTGCTGTSCGGCTGTGATGCGT"
c = Contig(query, subject, "Example")
qcopy = query.copy()
scopy = subject.copy()
c.reverse_direction()
assert c.query.start == qcopy.end
assert c.query.end == qcopy.start
assert c.subject.start == scopy.end
assert c.subject.end == scopy.start
assert c.query.sequence == qcopy.sequence[::-1]
assert c.subject.sequence == scopy.sequence[::-1]
def test_contig_constructor_with_context_error(query_subject_example):
"""This should raise a ContigError since the region lengths between
r1 and r2 are different"""
query, subject = query_subject_example
with pytest.raises(ContigError):
r1 = Region(1, 2, context=Context(100, True, start_index=1))
r2 = Region(1, 20, context=Context(100, True, start_index=1))
Contig(r1, r2, "Throws error")
def test_divide_contigs():
c1 = Context(10000, True)
c2 = Context(8000, True)
q1 = ContigRegion(1000, 3000, context=c1, forward=True)
s1 = ContigRegion(5000, 7000, context=c2, forward=True)
contig = Contig(q1, s1, "BLAST")
# not including self
divided_contigs = contig.divide_contig(
[1100, 1200],
[2000, 2200, 2300],
include_self=False)
assert len(divided_contigs) == 12
# including self
divided_contigs = contig.divide_contig(
[1100, 1200],
[2000, 2200, 2300],
include_self=True)
assert len(divided_contigs) == 13
def test_modify_subject_end(query_subject_example):
query, subject = query_subject_example
c = Contig(query, subject, "Example")
r = c.subject
meth = c.modify_subject_end
p1 = r.end+int(r.length * 0.9)
p2 = r.end+int(r.length * -0.9)
p3 = r.end+int(r.length * 200.1)
p4 = r.end+int(r.length * -200.1)
for p in [p1, p2]:
meth(p)
assert r.end == r.context.translate_pos(p)
for p in [p3, p4]:
with pytest.raises(RegionError):
meth(p)
def test_contig_schema():
c1 = Context(10000, True)
c2 = Context(5100, True)
q1 = ContigRegion(100, 300, context=c1, forward=True)
s1 = ContigRegion(500, 700, context=c2, forward=True)
contig1 = Contig(q1, s1, "BLAST", data1=5)
schema = schemas.ContigSchema()
data = schema.dump(contig1)
assert data['query']['start'] == 100
assert data['query']['end'] == 300
assert data['query']['context']['length'] == 10000
assert data['subject']['start'] == 500
assert data['subject']['end'] == 700
assert data['subject']['context']['length'] == 5100
assert data['quality']
assert data['metadata'] == {'data1': 5}
assert data['alignment_length'] == 201
def perfect_matches(self, rc=True):
"""
Pseudo-blast for finding perfect sequence matches (i.e. primers)
:param rc:
:return:
"""
# Get primer sequences (defined in db)
out, seqs, metadata = self.concate_db_to_fsa()
contig_container = ContigContainer()
# Get the query sequence
query_seq = open_sequence(self.query)[0].seq
query_seq_str = str(query_seq)
query_seq_str = re.sub('[nN]', '.', query_seq_str).lower()
fwd_matches = []
rev_matches = []
for seq in seqs:
seq_str = str(seq.seq)
try:
rc_seq_str = dna_reverse_complement(seq_str)
except KeyError:
continue
seq_str = re.sub('[nN]', '.', seq_str).lower()
rc_seq_str = re.sub('[nN]', '.', rc_seq_str).lower()
for match in re.finditer(seq_str, query_seq_str):
subject = ContigRegion(
seq.id,
ContigRegion.START_INDEX,
len(seq),
len(seq),
False,
True,
sequence=seq_str,
)
query = ContigRegion(
self.query,
match.start() + ContigRegion.START_INDEX,
match.end() + ContigRegion.START_INDEX - 1,
self.query_length,
self.query_circular,
True,
)
c = Contig(query, subject, Contig.TYPE_PRIMER)
contig_container.contigs.append(c)
for match in re.finditer(rc_seq_str, query_seq_str):
subject = ContigRegion(
seq.id,
len(seq),
ContigRegion.START_INDEX,
len(seq),
False,
False,
sequence=seq_str,
)
query = ContigRegion(
self.query,
match.start() + ContigRegion.START_INDEX,
match.end() + ContigRegion.START_INDEX - 1,
self.query_length,
self.query_circular,
True,
)
c = Contig(query, subject, Contig.TYPE_PRIMER)
contig_container.contigs.append(c)
return contig_container
def test_alignment_length(query_subject_example):
query, subject = query_subject_example
c = Contig(query, subject, "Example")
assert c.alignment_length == query.length
def test_invalid_region(query_subject_example):
query, subject = query_subject_example
query.start = 2
with pytest.raises(ContigError):
c = Contig(query, subject, "Example")
def test_contig_constructor(query_subject_example):
"""This tests constructing a contig from a subject and query"""
query, subject = query_subject_example
c = Contig(query, subject, "Example")
assert c.query == query
assert c.subject == subject
def test_fuse():
# Test fuse
c1 = Context(10000, True)
c2 = Context(5100, True)
q1 = ContigRegion(100, 300, context=c1, forward=True)
q2 = ContigRegion(301, 600, context=c1, forward=True)
q3 = ContigRegion(301, 600, context=c2, forward=True)
s1 = ContigRegion(500, 700, context=c2, forward=True)
s2 = ContigRegion(701, 1000, context=c2, forward=True)
s3 = ContigRegion(701, 1000, context=c2, forward=True)
contig1 = Contig(q1, s1, "BLAST")
contig2 = Contig(q2, s2, "BLAST")
contig1.fuse(contig2)
assert contig1.query.end == 600
# Raises error if one has sequence and other does not
contig1 = Contig(q1, s1, "BLAST")
contig2 = Contig(q2, s2, "BLAST")
a = "A"*contig1.query.length
b = "B"*contig1.subject.length
c = "C"*contig2.query.length
d = "D"*contig2.subject.length
contig1.query.sequence = a
contig1.subject.sequence = b
contig2.query.sequence = c
contig2.subject.sequence = d
contig1.fuse(contig2)
assert contig1.query.sequence == a + c
assert contig1.subject.sequence == b + d
# Raises error if one has sequence and other does not
contig1 = Contig(q1, s1, "BLAST")
contig2 = Contig(q2, s2, "BLAST")
contig1.query.sequence = "AGTCTGAGCTGTCGTGATAGTGCTGA"
contig1.subject.sequence = "AGTYAGYCHYAYHCYHYSCHYHAYCY"
with pytest.raises(ContigError):
contig1.fuse(contig2)
# Raise error if different contexts
contig1 = Contig(q1, s1, "BLAST")
contig2 = Contig(q2, s2, "BLAST")
contig3 = Contig(q3, s3, "BLAST")
with pytest.raises(ContigError):
contig3.fuse(contig2)