def test_24_ENSG00000163519_codon_aln(self):
"""
test kpic-gappy with codon alignment of mammalian sequences
usage: clipkit 24_ENSG00000163519_codon_aln.fasta -m kpic-gappy
"""
input_file = f"{here.parent}/samples/24_ENSG00000163519_codon_aln.fasta"
output_file = "output/24_ENSG00000163519_codon_aln.fasta.clipkit"
kwargs = dict(
input_file=input_file,
output_file=output_file,
input_file_format='fasta',
output_file_format='fasta',
complement=False,
gaps=0.9,
mode=TrimmingMode.kpic_gappy,
use_log=False,
)
execute(**kwargs)
with open(
f"{here.parent}/expected/24_ENSG00000163519_codon_aln.fasta_kpic_gappy",
"r") as expected:
expected_content = expected.read()
with open(output_file, "r") as out_file:
output_content = out_file.read()
assert expected_content == output_content
def test_EOG091N44M8_nt(self):
"""
test kpic-gappy with nucleotide alignment of Penicillium sequences
usage: clipkit EOG091N44M8_nt.fa -m kpic-gappy
"""
input_file = f"{here.parent}/samples/EOG091N44M8_nt.fa"
output_file = "output/EOG091N44M8_nt.fa.clipkit"
kwargs = dict(
input_file=input_file,
output_file=output_file,
input_file_format='fasta',
output_file_format='fasta',
complement=False,
gaps=0.9,
mode=TrimmingMode.kpic_gappy,
use_log=False,
)
execute(**kwargs)
with open(f"{here.parent}/expected/EOG091N44M8_nt.fa_kpic_gappy",
"r") as expected:
expected_content = expected.read()
with open(output_file, "r") as out_file:
output_content = out_file.read()
assert expected_content == output_content
def test_12_YIL115C_Anc_2_253_codon_aln(self):
"""
test gappy with codon alignment of yeast sequences
usage: clipkit 12_YIL115C_Anc_2.253_codon_aln.fasta -g 0.3
"""
input_file = f"{here.parent}/samples/12_YIL115C_Anc_2.253_codon_aln.fasta"
output_file = "output/12_YIL115C_Anc_2.253_codon_aln.fasta.clipkit"
in_file_format = 'fasta'
out_file_format = 'fasta'
kwargs = dict(
input_file=input_file,
output_file=output_file,
input_file_format='fasta',
output_file_format='fasta',
complement=False,
gaps=0.3,
mode=TrimmingMode.gappy,
use_log=False,
)
execute(**kwargs)
with open(
f"{here.parent}/expected/12_YIL115C_Anc_2.253_codon_aln.fasta_gappy_custom_gaps",
"r",
) as expected:
expected_content = expected.read()
with open(output_file, "r") as out_file:
output_content = out_file.read()
assert expected_content == output_content
def test_simple(self):
"""
usage: clipkit simple.fa -m kpic-gappy
"""
input_file = f"{here.parent}/samples/simple.fa"
output_file = "output/simpla.fa.TestKpiMode_test_simple.clipkit"
kwargs = dict(
input_file=input_file,
output_file=output_file,
input_file_format='fasta',
output_file_format='fasta',
complement=False,
gaps=0.9,
mode=TrimmingMode.kpic_gappy,
use_log=False,
)
execute(**kwargs)
with open(f"{here.parent}/expected/simple.fa_kpic_gappy",
"r") as expected:
expected_content = expected.read()
with open(output_file, "r") as out_file:
output_content = out_file.read()
assert expected_content == output_content
def test_simple_no_change(self):
"""
test gappy where no changes are expected in the resulting
output alignment.
usage: clipkit simple.fa
"""
input_file = f"{here.parent}/samples/simple.fa"
output_file = "output/simpla.fa.clipkit"
kwargs = dict(
input_file=input_file,
output_file=output_file,
input_file_format='fasta',
output_file_format='fasta',
complement=False,
gaps=0.9,
mode=TrimmingMode.gappy,
use_log=False,
)
execute(**kwargs)
with open(input_file, "r") as expected:
expected_content = expected.read()
with open(output_file, "r") as out_file:
output_content = out_file.read()
assert expected_content == output_content
def test_simple(self):
"""
test gappy with a custom gaps parameter
usage: clipkit simple.fa -g 0.2
"""
input_file = f"{here.parent}/samples/simple.fa"
output_file = "output/simpla.fa.clipkit"
in_file_format = 'fasta'
out_file_format = 'fasta'
kwargs = dict(
input_file=input_file,
output_file=output_file,
input_file_format='fasta',
output_file_format='fasta',
complement=False,
gaps=0.2,
mode=TrimmingMode.gappy,
use_log=False,
)
execute(**kwargs)
with open(f"{here.parent}/expected/simple.fa_gappy_gaps_set_to_0.2",
"r") as expected:
expected_content = expected.read()
with open(output_file, "r") as out_file:
output_content = out_file.read()
assert expected_content == output_content
def test_simple_long_description(self):
"""
test output in clustal format
usage: clipkit simple_long_description.fa -l
"""
input_file = f"{here.parent}/samples/simple_long_description.fa"
output_file = "output/simple_long_description.fa.clipkit"
kwargs = dict(
input_file=input_file,
output_file=output_file,
input_file_format='fasta',
output_file_format='fasta',
complement=False,
gaps=0.9,
mode=TrimmingMode.gappy,
use_log=True,
)
execute(**kwargs)
with open(
f"{here.parent}/expected/simple_long_description.fa.clipkit.log",
"r") as expected:
expected_content = expected.read()
with open(f"{output_file}.log", "r") as out_file:
output_content = out_file.read()
assert expected_content == output_content
def test_24_ENSG00000163519_aa_aln(self):
"""
test gappy with amino acid alignment of mammalian sequences
usage: clipkit 24_ENSG00000163519_aa_aln.fasta
"""
input_file = f"{here.parent}/samples/24_ENSG00000163519_aa_aln.fasta"
output_file = "output/24_ENSG00000163519_aa_aln.fasta.clipkit"
in_file_format = 'fasta'
out_file_format = 'fasta'
kwargs = dict(
input_file=input_file,
output_file=output_file,
input_file_format='fasta',
output_file_format='fasta',
complement=False,
gaps=0.9583,
mode=TrimmingMode.smart_gap,
use_log=False,
)
execute(**kwargs)
with open(
f"{here.parent}/expected/24_ENSG00000163519_aa_aln.clipkit_smart_gaps",
"r") as expected:
expected_content = expected.read()
with open(output_file, "r") as out_file:
output_content = out_file.read()
assert expected_content == output_content
def test_12_YIL115C_Anc_2_253_aa_aln_complement(self):
"""
test complementary output file for amino acid yeast sequences
usage: clipkit 12_YIL115C_Anc_2.253_aa_aln.fasta -c
"""
output_file = "output/12_YIL115C_Anc_2.253_aa_aln.fasta_gappy"
complement_out_file = f"{output_file}.complement"
kwargs = dict(
input_file=
f"{here.parent}/samples/12_YIL115C_Anc_2.253_aa_aln.fasta",
output_file=output_file,
input_file_format='fasta',
output_file_format='fasta',
complement=True,
gaps=0.9,
mode=TrimmingMode.gappy,
use_log=False,
)
execute(**kwargs)
with open(
f"{here.parent}/expected/12_YIL115C_Anc_2.253_aa_aln.fasta_gappy.complement",
"r",
) as expected:
expected_content = expected.read()
with open(complement_out_file, "r") as out_file:
output_content = out_file.read()
assert expected_content == output_content
def test_EOG091N44M8_aa_complement(self):
"""
test complementary output file for amino acid Penicillium sequences
usage: clipkit EOG091N44M8_aa.fa -c
"""
output_file = "output/EOG091N44M8_aa.fa_gappy"
complement_out_file = f"{output_file}.complement"
kwargs = dict(
input_file=f"{here.parent}/samples/EOG091N44M8_aa.fa",
output_file=output_file,
input_file_format='fasta',
output_file_format='fasta',
complement=True,
gaps=0.9,
mode=TrimmingMode.gappy,
use_log=False,
)
execute(**kwargs)
with open(f"{here.parent}/expected/EOG091N44M8_aa.fa_gappy.complement",
"r") as expected:
expected_content = expected.read()
with open(complement_out_file, "r") as out_file:
output_content = out_file.read()
assert expected_content == output_content
def test_simple_complement(self):
"""
test complementary output file with a simple case
usage: clipkit simple.fa -c
"""
output_file = "output/simple.fa_gappy"
complement_out_file = f"{output_file}.complement"
kwargs = dict(
input_file=f"{here.parent}/samples/simple.fa",
output_file=output_file,
input_file_format='fasta',
output_file_format='fasta',
complement=True,
gaps=0.9,
mode=TrimmingMode.gappy,
use_log=False,
)
execute(**kwargs)
with open(f"{here.parent}/expected/simple.fa_gappy.complement",
"r") as expected:
expected_content = expected.read()
with open(complement_out_file, "r") as out_file:
output_content = out_file.read()
assert expected_content == output_content
def test_simple_no_change(self):
"""
usage: clipkit simple_long_description.fa -m kpic-smart-gap
"""
input_file = f"{here.parent}/samples/simple_long_description.fa"
output_file = "output/simple_long_description.fa_kpic_smart_gaps"
kwargs = dict(
input_file=input_file,
output_file=output_file,
input_file_format='fasta',
output_file_format='fasta',
complement=False,
gaps=0.8,
mode=TrimmingMode.kpic_smart_gap,
use_log=False,
)
execute(**kwargs)
with open(
f"{here.parent}/expected/simple_long_description.fa_kpic_smart_gaps",
"r") as expected:
expected_content = expected.read()
with open(output_file, "r") as out_file:
output_content = out_file.read()
assert expected_content == output_content
def test_EOG091N44M8_aa(self):
"""
test gappy with amino acid alignment of Penicillium sequences
usage: clipkit EOG091N44M8_aa.fa -m kpic-smart-gap
"""
input_file = f"{here.parent}/samples/EOG091N44M8_aa.fa"
output_file = "output/EOG091N44M8_aa.fa.clipkit"
in_file_format = 'fasta'
out_file_format = 'fasta'
kwargs = dict(
input_file=input_file,
output_file=output_file,
input_file_format='fasta',
output_file_format='fasta',
complement=False,
gaps=0.8803,
mode=TrimmingMode.kpic_smart_gap,
use_log=False,
)
execute(**kwargs)
with open(
f"{here.parent}/expected/EOG091N44M8_aa.clipkit_kpic_smart_gaps",
"r") as expected:
expected_content = expected.read()
with open(output_file, "r") as out_file:
output_content = out_file.read()
assert expected_content == output_content
def test_EOG092C0CZK_aa(self):
"""
test kpic-gappy with amino alignment of fungal sequences
usage: clipkit EOG092C0CZK_aa_aln.fasta -m kpic-gappy
"""
input_file = f"{here.parent}/samples/EOG092C0CZK_aa_aln.fasta"
output_file = "output/EOG092C0CZK_aa_aln.fasta.clipkitc"
kwargs = dict(
input_file=input_file,
output_file=output_file,
input_file_format='fasta',
output_file_format='fasta',
complement=False,
gaps=0.9,
mode=TrimmingMode.kpic_gappy,
use_log=False,
)
execute(**kwargs)
with open(
f"{here.parent}/expected/EOG092C0CZK_aa_aln.fasta_kpic_gappy",
"r") as expected:
expected_content = expected.read()
with open(output_file, "r") as out_file:
output_content = out_file.read()
assert expected_content == output_content
def test_12_YIL115C_Anc_2_253_codon_aln(self):
"""
test output in clustal format
usage: clipkit 12_YIL115C_Anc_2.253_codon_aln.fasta -l
"""
input_file = f"{here.parent}/samples/12_YIL115C_Anc_2.253_codon_aln.fasta"
output_file = "output/12_YIL115C_Anc_2.253_codon_aln.fasta.clipkit"
kwargs = dict(
input_file=input_file,
output_file=output_file,
input_file_format='fasta',
output_file_format='fasta',
complement=False,
gaps=0.9,
mode=TrimmingMode.gappy,
use_log=True,
)
execute(**kwargs)
with open(
f"{here.parent}/expected/12_YIL115C_Anc_2.253_codon_aln.fasta.clipkit.log",
"r",
) as expected:
expected_content = expected.read()
with open(f"{output_file}.log", "r") as out_file:
output_content = out_file.read()
assert expected_content == output_content
def test_12_YIL115C_Anc_2_253_aa_aln(self):
"""
test kpic-gappy with amino acid alignment of yeast sequences
usage: clipkit 12_YIL115C_Anc_2.253_aa_aln.fasta -m kpic-gappy
"""
input_file = f"{here.parent}/samples/12_YIL115C_Anc_2.253_aa_aln.fasta"
output_file = "output/12_YIL115C_Anc_2.253_aa_aln.fasta.clipkit"
kwargs = dict(
input_file=input_file,
output_file=output_file,
input_file_format='fasta',
output_file_format='fasta',
complement=False,
gaps=0.9,
mode=TrimmingMode.kpic_gappy,
use_log=False,
)
execute(**kwargs)
with open(
f"{here.parent}/expected/12_YIL115C_Anc_2.253_aa_aln.fasta_kpic_gappy",
"r") as expected:
expected_content = expected.read()
with open(output_file, "r") as out_file:
output_content = out_file.read()
assert expected_content == output_content
def test_phylip(self):
"""
test output in phylip format
usage: clipkit simple.fa -of phylip
"""
input_file = f"{here.parent}/samples/simple.fa"
output_file = "output/simple.phylip"
kwargs = dict(
input_file=input_file,
output_file=output_file,
input_file_format='fasta',
output_file_format='phylip',
complement=False,
gaps=0.9,
mode=TrimmingMode.gappy,
use_log=False,
)
execute(**kwargs)
with open(f"{here.parent}/expected/simple.phylip", "r") as expected:
expected_content = expected.read()
with open(output_file, "r") as out_file:
output_content = out_file.read()
assert expected_content == output_content