def test_join(self):
"""
Test BioSeqs.join() method.
"""
infile1 = 'Fasta/f001.fasta'
infile2 = 'Phylip/f003.phylip'
self.assertTrue(os.path.isfile(infile1))
self.assertTrue(os.path.isfile(infile2))
seq_db = BioSeqs.from_seqfile(infile1, 'fasta')
extra_db = BioSeqs.from_seqfile(infile2, 'phylip')
seq_db.join(extra_db)
# Check the sequence data
indict1 = SeqIO.to_dict(SeqIO.parse(infile1, 'fasta'))
indict2 = SeqIO.to_dict(SeqIO.parse(infile2, 'phylip'))
self.assertEqual(len(indict1) + len(indict2), len(seq_db))
for key, value in viewitems(indict1):
self.assertEqual(str(value.seq), str(seq_db.data[key].seq))
for key, value in viewitems(indict2):
self.assertEqual(str(value.seq), str(seq_db.data[key].seq))
# Check the report information
self.assertIn('local', seq_db._report[0][1])
self.assertIn('Tests/Fasta/f001.fasta', seq_db._report[0][2])
self.assertIn('fasta', seq_db._report[0][3])
self.assertIn('local', seq_db._report[1][1])
self.assertIn('Tests/Phylip/f003.phylip', seq_db._report[1][2])
self.assertIn('phylip', seq_db._report[1][3])
def test_join ( self ) :
"""
Test BioSeqs.join() method.
"""
infile1 = 'Fasta/f001.fasta'
infile2 = 'Phylip/f003.phylip'
self.assertTrue(os.path.isfile(infile1))
self.assertTrue(os.path.isfile(infile2))
seq_db = BioSeqs.from_seqfile(infile1, 'fasta')
extra_db = BioSeqs.from_seqfile(infile2, 'phylip')
seq_db.join(extra_db)
# Check the sequence data
indict1 = SeqIO.to_dict(SeqIO.parse(infile1, 'fasta'))
indict2 = SeqIO.to_dict(SeqIO.parse(infile2, 'phylip'))
self.assertEqual(len(indict1) + len(indict2), len(seq_db))
for key, value in viewitems(indict1) :
self.assertEqual(str(value.seq), str(seq_db.data[key].seq))
for key, value in viewitems(indict2) :
self.assertEqual(str(value.seq), str(seq_db.data[key].seq))
# Check the report information
self.assertIn('local', seq_db._report[0][1])
self.assertIn('Tests/Fasta/f001.fasta', seq_db._report[0][2])
self.assertIn('fasta', seq_db._report[0][3])
self.assertIn('local', seq_db._report[1][1])
self.assertIn('Tests/Phylip/f003.phylip', seq_db._report[1][2])
self.assertIn('phylip', seq_db._report[1][3])
def test_seqfile_source(self):
"""
Test BioSeqs.from_seqfile() and BioSeqs.write() methods.
"""
infile = 'Fasta/f001.fasta'
self.assertTrue(os.path.isfile(infile))
seq_db = BioSeqs.from_seqfile(infile, 'fasta')
outfile = 'tmp_test.gb'
outrepfile = 'tmp_test.rep'
self.files_to_clean.add(outfile)
self.files_to_clean.add(outrepfile)
seq_db.write(outfile)
self.assertTrue(os.path.isfile(outfile))
# Check the content of both sequence files
indict = SeqIO.to_dict(SeqIO.parse(infile, 'fasta'))
outdict = SeqIO.to_dict(SeqIO.parse(outfile, 'gb'))
self.assertEqual(len(indict), len(outdict))
for key, value in viewitems(indict):
self.assertEqual(str(value.seq), str(outdict[key].seq))
# Check the content of the report file
with open(outrepfile, 'r') as repfile:
for line in repfile.readlines():
self.assertTrue(
('Num. sequences: 50' in line) or ('History:' in line)
or (bool(
re.match(
r"""\d\d\d\d/\d\d/\d\d\ \d\d:\d\d:\d\d[ ]+
local[ ]+.*Tests/Fasta/f001\.fasta
[ ]+fasta""", line, re.VERBOSE))))
def test_seqfile_source ( self ) :
"""
Test BioSeqs.from_seqfile() and BioSeqs.write() methods.
"""
infile = 'Fasta/f001.fasta'
self.assertTrue(os.path.isfile(infile))
seq_db = BioSeqs.from_seqfile(infile, 'fasta')
outfile = 'tmp_test.gb'
outrepfile = 'tmp_test.rep'
self.files_to_clean.add(outfile)
self.files_to_clean.add(outrepfile)
seq_db.write(outfile)
self.assertTrue(os.path.isfile(outfile))
# Check the content of both sequence files
indict = SeqIO.to_dict(SeqIO.parse(infile, 'fasta'))
outdict = SeqIO.to_dict(SeqIO.parse(outfile, 'gb'))
self.assertEqual(len(indict), len(outdict))
for key, value in viewitems(indict) :
self.assertEqual(str(value.seq), str(outdict[key].seq))
# Check the content of the report file
with open(outrepfile, 'r') as repfile :
for line in repfile.readlines() :
self.assertTrue(('Num. sequences: 50' in line) or
('History:' in line) or
(bool(re.match(r"""\d\d\d\d/\d\d/\d\d\ \d\d:\d\d:\d\d[ ]+
local[ ]+.*Tests/Fasta/f001\.fasta
[ ]+fasta""", line, re.VERBOSE))))
def test_statistics(self):
"""
Test BioSeqs.statistics() method.
"""
infile = 'Fasta/f001.fasta'
self.assertTrue(os.path.isfile(infile))
seq_db = BioSeqs.from_seqfile(infile, 'fasta')
nseqs, mean, stdev, minimum, maximum = seq_db.statistics()
# Check the resultant values
self.assertEqual(nseqs, len(seq_db))
self.assertEqual(mean, 579.02)
self.assertEqual(stdev, 0.14)
self.assertEqual(minimum, 579)
self.assertEqual(maximum, 580)
def test_statistics ( self ) :
"""
Test BioSeqs.statistics() method.
"""
infile = 'Fasta/f001.fasta'
self.assertTrue(os.path.isfile(infile))
seq_db = BioSeqs.from_seqfile(infile, 'fasta')
nseqs, mean, stdev, minimum, maximum = seq_db.statistics()
# Check the resultant values
self.assertEqual(nseqs, len(seq_db))
self.assertEqual(mean, 579.02)
self.assertEqual(stdev, 0.14)
self.assertEqual(minimum, 579)
self.assertEqual(maximum, 580)
def test_str ( self ) :
"""
Test str() property.
"""
infile = 'Fasta/f001.fasta'
self.assertTrue(os.path.isfile(infile))
seq_db = BioSeqs.from_seqfile(infile, 'fasta')
outstr = str(seq_db)
# Check the output string
lines = outstr.split('\n')
self.assertTrue(bool(re.match(r"""DB:\ \{clpA_\d+,\ clpA_\d+,\ clpA_\d+,
\ clpA_\d+,\ clpA_\d+,\.\.\.\}""", lines[0], re.VERBOSE)))
self.assertEqual(lines[1], 'Num. sequences: 50')
self.assertEqual(lines[2], 'History:')
self.assertTrue(bool(re.match(r"""\d\d\d\d/\d\d/\d\d\ \d\d:\d\d:\d\d[ ]+
local[ ]+.*Tests/Fasta/f001\.fasta[ ]+fasta""", lines[3],
re.VERBOSE)))
def test_str(self):
"""
Test str() property.
"""
infile = 'Fasta/f001.fasta'
self.assertTrue(os.path.isfile(infile))
seq_db = BioSeqs.from_seqfile(infile, 'fasta')
outstr = str(seq_db)
# Check the output string
lines = outstr.split('\n')
self.assertTrue(
bool(
re.match(
r"""DB:\ \{clpA_\d+,\ clpA_\d+,\ clpA_\d+,
\ clpA_\d+,\ clpA_\d+,\.\.\.\}""", lines[0], re.VERBOSE)))
self.assertEqual(lines[1], 'Num. sequences: 50')
self.assertEqual(lines[2], 'History:')
self.assertTrue(
bool(
re.match(
r"""\d\d\d\d/\d\d/\d\d\ \d\d:\d\d:\d\d[ ]+
local[ ]+.*Tests/Fasta/f001\.fasta[ ]+fasta""", lines[3],
re.VERBOSE)))
