def test_top_sequence_3(self):
sequence_file = SequenceFile('test')
sequence1 = Sequence('foo', 'AAAAA')
sequence2 = Sequence('bar', 'CCCCC')
sequence_file.add(sequence1)
sequence_file.add(sequence2)
self.assertEqual(sequence1, sequence_file.top_sequence)
def test_remark_3(self):
sequence_file = SequenceFile("test")
sequence_file.remark = "Hello"
sequence_file.remark = "5"
sequence_file.remark = "World"
sequence_file.remark = "!"
self.assertEqual(["Hello", "5", "World", "!"], sequence_file.remark)
def read(self, f_handle, f_id="a2m"):
"""Read a sequence file
Parameters
----------
f_handle
Open file handle [read permissions]
f_id : str, optional
Unique sequence file identifier
Returns
-------
:obj:`~conkit.core.sequencefile.SequenceFile`
"""
hierarchy = SequenceFile(f_id)
for i, line in enumerate(f_handle):
line = line.strip()
if line:
for j, char in enumerate(line):
if char.isalpha() or char == "-":
continue
indicator = ["-"] * len(line)
indicator[j] = "^"
msg = "Unknown character in line {0}:{1}{1}{2}{1}{3}"
msg = msg.format(i + 1, "\n", line, "".join(indicator))
raise ValueError(msg)
sequence = Sequence("seq_{}".format(i), line)
hierarchy.add(sequence)
return hierarchy
def test_remark_3(self):
sequence_file = SequenceFile('test')
sequence_file.remark = 'Hello'
sequence_file.remark = '5'
sequence_file.remark = 'World'
sequence_file.remark = '!'
self.assertEqual(['Hello', '5', 'World', '!'], sequence_file.remark)
def test_top_sequence_3(self):
sequence_file = SequenceFile("test")
sequence1 = Sequence("foo", "AAAAA")
sequence2 = Sequence("bar", "CCCCC")
sequence_file.add(sequence1)
sequence_file.add(sequence2)
self.assertEqual(sequence1, sequence_file.top_sequence)
def test_sort_4(self):
sequence_file = SequenceFile("test")
for seq in [Sequence("foo", "AAAAA"), Sequence("bar", "DDDDD"), Sequence("doe", "CCCCC")]:
sequence_file.add(seq)
sequence_file_sorted = sequence_file.sort("seq", reverse=True, inplace=True)
self.assertEqual(["bar", "doe", "foo"], [s.id for s in sequence_file_sorted])
self.assertEqual(["DDDDD", "CCCCC", "AAAAA"], [s.seq for s in sequence_file_sorted])
self.assertEqual(sequence_file, sequence_file_sorted)
def test_trim_4(self):
sequence_file = SequenceFile("test")
for seq in [Sequence("foo", "ACDEF"), Sequence("bar", "CDEFG"), Sequence("doe", "DEFGH")]:
sequence_file.add(seq)
sequence_file_trimmed = sequence_file.trim(2, 3)
self.assertEqual(["foo", "bar", "doe"], [s.id for s in sequence_file_trimmed])
self.assertEqual(["CD", "DE", "EF"], [s.seq for s in sequence_file_trimmed])
self.assertNotEqual(sequence_file, sequence_file_trimmed)
def test_remark_5(self):
sequence_file = SequenceFile("test")
sequence_file.remark = "hello"
sequence = Sequence("foo", "GSMFTPK")
sequence.remark = "bar"
sequence_file.add(sequence)
self.assertEqual(["hello"], sequence_file.remark)
self.assertEqual(["bar"], sequence_file[0].remark)
def test_ascii_matrix_1(self):
sequence_file = SequenceFile("test")
for seq in [Sequence("foo", "AAAAAA"), Sequence("bar", "-CC-C-"), Sequence("doe", "DDDDDD")]:
sequence_file.add(seq)
matrix = sequence_file.ascii_matrix
self.assertEqual([65, 65, 65, 65, 65, 65], list(matrix)[0])
self.assertEqual([45, 67, 67, 45, 67, 45], list(matrix)[1])
self.assertEqual([68, 68, 68, 68, 68, 68], list(matrix)[2])
def test_remark_5(self):
sequence_file = SequenceFile('test')
sequence_file.remark = 'hello'
sequence = Sequence('foo', 'GSMFTPK')
sequence.remark = 'bar'
sequence_file.add(sequence)
self.assertEqual(['hello'], sequence_file.remark)
self.assertEqual(['bar'], sequence_file[0].remark)
def test_trim_2(self):
sequence_file = SequenceFile("test")
for seq in [Sequence("foo", "AAAAA"), Sequence("bar", "CCCCC"), Sequence("doe", "DDDDD")]:
sequence_file.add(seq)
sequence_file_trimmed = sequence_file.trim(3, 5)
self.assertEqual(["foo", "bar", "doe"], [s.id for s in sequence_file_trimmed])
self.assertEqual(["AAA", "CCC", "DDD"], [s.seq for s in sequence_file_trimmed])
self.assertNotEqual(sequence_file, sequence_file_trimmed)
def read(self, f_handle, f_id='fasta'):
"""Read a sequence file
Parameters
----------
f_handle
Open file handle [read permissions]
f_id : str, optional
Unique sequence file identifier
Returns
-------
:obj:`SequenceFile <conkit.core.sequencefile.SequenceFile>`
"""
# Create a new sequence file instance
hierarchy = SequenceFile(f_id)
# Read any possible comments and store in file remarks
while True:
line = f_handle.readline().rstrip()
if not line:
continue
elif line.startswith('#'):
hierarchy.remark = line[1:]
elif line.startswith('>'):
break
# Read the sequence record(s) and store them
while True:
if not line.startswith('>'):
raise ValueError("Fasta record needs to start with '>'")
id = line[1:] # Header without '>'
chunks = []
line = f_handle.readline().rstrip()
while True:
if not line:
break
elif line.startswith('>'):
break
chunks.append(line)
line = f_handle.readline().rstrip()
_seq_string = "".join(chunks) # Sequence from chunks
# Create the sequence record instance
sequence_entry = Sequence(id, _seq_string)
# Store the sequence in the file
hierarchy.add(sequence_entry)
if not line:
break
return hierarchy
def test_diversity_2(self):
sequence_file = SequenceFile('test')
for seq in [
Sequence('foo', 'AAAAAA'),
Sequence('bar', 'AAAAAA'),
Sequence('doe', 'AAAAAA')
]:
sequence_file.add(seq)
self.assertEqual(0.289, sequence_file.diversity)
def read(self, f_handle, f_id="fasta"):
"""Read a sequence file
Parameters
----------
f_handle
Open file handle [read permissions]
f_id : str, optional
Unique sequence file identifier
Returns
-------
:obj:`~conkit.core.sequencefile.SequenceFile`
Raises
------
:exc:`ValueError`
FASTA record needs to start with >
"""
hierarchy = SequenceFile(f_id)
while True:
line = f_handle.readline().rstrip()
if not line:
continue
elif line.startswith("#"):
hierarchy.remark = line[1:]
elif line.startswith(">"):
break
while True:
if not line.startswith(">"):
raise ValueError("Fasta record needs to start with '>'")
id = line[1:] # Header without '>'
chunks = []
line = f_handle.readline().rstrip()
while True:
if not line:
break
elif line.startswith(">"):
break
chunks.append(line)
line = f_handle.readline().rstrip()
_seq_string = "".join(chunks) # Sequence from chunks
sequence_entry = Sequence(id, _seq_string)
hierarchy.add(sequence_entry)
if not line:
break
return hierarchy
def test_get_frequency_1(self):
sequence_file = SequenceFile('test')
for s in [
Sequence('foo', 'AAAAAAA'),
Sequence('bar', 'A-AAAA-'),
Sequence('cho', '--AAA--')
]:
sequence_file.add(s)
self.assertEqual([1, 2, 0, 0, 0, 1, 2],
sequence_file.get_frequency("X"))
def test_diversity_4(self):
sequence_file = SequenceFile('test')
for seq in [
Sequence('foo', 'AAAAAA'),
Sequence('bar', 'B'),
Sequence('doe', 'CCC')
]:
sequence_file.add(seq)
with self.assertRaises(ValueError):
sequence_file.diversity
def test_filter_6(self):
sequence_file = SequenceFile('test')
for seq in [
Sequence('foo', 'AAAAAA'),
Sequence('bar', 'AACCCC'),
Sequence('doe', 'DDDDDD')
]:
sequence_file.add(seq)
filtered = sequence_file.filter(min_id=0.1, max_id=0.9)
self.assertEqual(['foo', 'bar'], [s.id for s in filtered])
def test_get_frequency_2(self):
sequence_file = SequenceFile('test')
for s in [
Sequence('foo', '-------'),
Sequence('bar', '-------'),
Sequence('cho', '-------')
]:
sequence_file.add(s)
self.assertEqual([3, 3, 3, 3, 3, 3, 3],
sequence_file.get_frequency("-"))
def test_get_frequency_3(self):
sequence_file = SequenceFile('test')
for s in [
Sequence('foo', 'AAAAAAA'),
Sequence('bar', 'AAAAAAA'),
Sequence('cho', 'AAAAAAA')
]:
sequence_file.add(s)
self.assertEqual([0, 0, 0, 0, 0, 0, 0],
sequence_file.get_frequency("X"))
def test_get_weights_5(self):
sequence_file = SequenceFile("test")
for s in [
Sequence("foo", "AAAAAAA"),
Sequence("bar", "AA-ACA-"),
Sequence("cho", "C-CAA--"),
Sequence("baz", "CCCCCCC"),
]:
sequence_file.add(s)
weights = sequence_file.get_weights(identity=0.6)
self.assertEqual(weights, [1.0, 1.0, 1.0, 1.0])
def test_get_weights_5(self):
sequence_file = SequenceFile('test')
for s in [
Sequence('foo', 'AAAAAAA'),
Sequence('bar', 'AA-ACA-'),
Sequence('cho', 'C-CAA--'),
Sequence('baz', 'CCCCCCC')
]:
sequence_file.add(s)
weights = sequence_file.get_weights(identity=0.6)
self.assertEqual(weights, [1.0, 1.0, 1.0, 1.0])
def test_get_weights_1(self):
sequence_file = SequenceFile('test')
for s in [
Sequence('foo', 'AAAAAAA'),
Sequence('bar', 'AAAAAAA'),
Sequence('cho', 'AAAAAAA'),
Sequence('baz', 'AAAAAAA')
]:
sequence_file.add(s)
weights = sequence_file.get_weights(identity=0.7)
self.assertEqual(weights, [0.25, 0.25, 0.25, 0.25])
def test_get_weights_2(self):
sequence_file = SequenceFile("test")
for s in [
Sequence("foo", "AAAAAAA"),
Sequence("bar", "AAAAAAA"),
Sequence("cho", "AAAAAAA"),
Sequence("baz", "CCCCCCC"),
]:
sequence_file.add(s)
weights = sequence_file.get_weights(identity=0.7)
self.assertEqual(weights, [0.3333333333333333, 0.3333333333333333, 0.3333333333333333, 1.0])
def test_calculate_weights_5(self):
sequence_file = SequenceFile('test')
for s in [
Sequence('foo', 'AAAAAAA'),
Sequence('bar', 'AA-ABA-'),
Sequence('cho', 'B-BAA--'),
Sequence('baz', 'BBBBBBB')
]:
sequence_file.add(s)
weights = sequence_file.calculate_weights(identity=0.6)
self.assertEqual(weights, [1.0, 1.0, 1.0, 1.0])
def test_get_weights_6(self):
sequence_file = SequenceFile("test")
for s in [
Sequence("foo", "AAAAAAA"),
Sequence("bar", "AA-ACA-"),
Sequence("cho", "AAADCAA"),
Sequence("doo", "C-CAA--"),
Sequence("miu", "CCCCCCC"),
Sequence("nop", "AAAAAAB"),
]:
sequence_file.add(s)
self.assertEqual(5, sequence_file.meff)
def test_calculate_weights_6(self):
sequence_file = SequenceFile('test')
for s in [
Sequence('foo', 'AAAAAAA'),
Sequence('bar', 'AA-ABA-'),
Sequence('cho', 'AAACBAA'),
Sequence('doo', 'B-BAA--'),
Sequence('miu', 'BBBBBBB'),
Sequence('nop', 'AAAAAAB')
]:
sequence_file.add(s)
self.assertEqual(5, sequence_file.neff)
def test_calculate_freq_3(self):
sequence_file = SequenceFile('test')
for s in [
Sequence('foo', 'AAAAAAA'),
Sequence('bar', 'AAAAAAA'),
Sequence('cho', 'AAAAAAA')
]:
sequence_file.add(s)
calculated_freqs = [
round(i, 6) for i in sequence_file.calculate_freq()
]
self.assertEqual([1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0], calculated_freqs)
def test_get_weights_6(self):
sequence_file = SequenceFile('test')
for s in [
Sequence('foo', 'AAAAAAA'),
Sequence('bar', 'AA-ACA-'),
Sequence('cho', 'AAADCAA'),
Sequence('doo', 'C-CAA--'),
Sequence('miu', 'CCCCCCC'),
Sequence('nop', 'AAAAAAB')
]:
sequence_file.add(s)
self.assertEqual(5, sequence_file.meff)
def test_ascii_matrix_1(self):
sequence_file = SequenceFile('test')
for seq in [
Sequence('foo', 'AAAAAA'),
Sequence('bar', '-CC-C-'),
Sequence('doe', 'DDDDDD')
]:
sequence_file.add(seq)
matrix = sequence_file.ascii_matrix
self.assertEqual([65, 65, 65, 65, 65, 65], list(matrix)[0])
self.assertEqual([45, 67, 67, 45, 67, 45], list(matrix)[1])
self.assertEqual([68, 68, 68, 68, 68, 68], list(matrix)[2])
def test_get_weights_6(self):
sequence_file = SequenceFile("test")
for s in [
Sequence("foo", "AAAAAAA"),
Sequence("bar", "AA-ACA-"),
Sequence("cho", "AAADCAA"),
Sequence("doo", "C-CAA--"),
Sequence("miu", "CCCCCCC"),
Sequence("nop", "AAAAAAB"),
]:
sequence_file.add(s)
weights = sequence_file.get_weights(identity=0.6)
self.assertEqual(weights, [0.3333333333333333, 1.0, 0.5, 1.0, 1.0, 0.5])