def test_equivalent_different_deflines_ignore(tmpdir, capsys):
"""Verify the correct message when different sequence descriptions, but deflines are ignored."""
seq_strings = ['A' * 1]
path1 = write_fasta(seq_strings, tmpdir, "file1", '1')
seq_strings = ['A' * 1]
path2 = write_fasta(seq_strings, tmpdir, "file2", '2')
fastatools.equivalent(path1, path2, ignore_defline=True)
captured = capsys.readouterr()
assert (captured.out == "Equivalent\n")
def test_equivalent_different_contents(tmpdir, capsys):
"""Verify the correct message when different sequence contents."""
seq_strings = ['A' * 10]
path1 = write_fasta(seq_strings, tmpdir, "file1")
seq_strings = ['T' * 10]
path2 = write_fasta(seq_strings, tmpdir, "file2")
fastatools.equivalent(path1, path2, ignore_defline=False)
captured = capsys.readouterr()
assert (captured.out ==
'Not equivalent -- sequence 1 has different contents.\n')
def test_equivalent_different_num_sequences(tmpdir, capsys):
"""Verify the correct message when different number of sequences."""
seq_strings = ['A' * 1, 'T' * 10, 'C' * 100]
path1 = write_fasta(seq_strings, tmpdir, "A1.T10.C100")
seq_strings = ['A' * 1, 'T' * 10, 'C' * 100, 'G' * 1000]
path2 = write_fasta(seq_strings, tmpdir, "A2.T20.C200.G2000")
fastatools.equivalent(path1, path2)
captured = capsys.readouterr()
assert (
captured.out ==
"Not equivalent -- the number of sequences is different (3 and 4).\n")
def test_equivalent_different_deflines(tmpdir, capsys):
"""Verify the correct message when different sequence descriptions."""
seq_strings = ['A' * 1]
path1 = write_fasta(seq_strings, tmpdir, "file1", '1')
seq_strings = ['A' * 1]
path2 = write_fasta(seq_strings, tmpdir, "file2", '2')
fastatools.equivalent(path1, path2, ignore_defline=False)
captured = capsys.readouterr()
assert (
captured.out ==
'Not equivalent -- sequence 1 has different descriptions ("Id11 Description11" and "Id21 Description21").\n'
)
def test_equivalent_ignore_order(tmpdir, capsys):
"""Verify the sequences can be in different order"""
seq_strings = ['A' * 10, 'T' * 100]
path1 = write_fasta(seq_strings, tmpdir, "file1")
seq_strings = ['T' * 100, 'A' * 10]
path2 = write_fasta(seq_strings, tmpdir, "file2")
fastatools.equivalent(path1,
path2,
ignore_defline=True,
enforce_order=False)
captured = capsys.readouterr()
assert (captured.out == "Equivalent\n")
def test_equivalent_enforce_order(tmpdir, capsys):
"""Verify the sequences must be in the same order"""
seq_strings = ['A' * 10, 'T' * 100]
path1 = write_fasta(seq_strings, tmpdir, "file1")
seq_strings = ['T' * 100, 'A' * 10]
path2 = write_fasta(seq_strings, tmpdir, "file2")
fastatools.equivalent(path1,
path2,
ignore_defline=True,
enforce_order=True)
captured = capsys.readouterr()
assert (
captured.out ==
'Not equivalent -- sequence 1 has different lengths (10 and 100).\n'
'Not equivalent -- sequence 2 has different lengths (100 and 10).\n')
def equivalent_command(args):
"""Determine if two sequences are equivalent, ignoring the line lengths and uppercase / lowercase characters.
Parameters
----------
args : Namespace
Command line arguments stored as attributes of a Namespace, usually
parsed from sys.argv
"""
return fastatools.equivalent(args.fasta_path1,
args.fasta_path2,
ignore_defline=args.ignore_defline,
enforce_order=args.enforce_order)