def test_simple_clustal_strict(self):
"""Simple muscle call using strict Clustal output"""
input_file = "Fasta/f002"
self.assertTrue(os.path.isfile(input_file))
records = list(SeqIO.parse(input_file,"fasta"))
records.sort(key = lambda rec: rec.id)
#Prepare the command...
cmdline = MuscleCommandline(muscle_exe)
cmdline.set_parameter("in", input_file)
#Use clustal output (with a CLUSTAL header)
cmdline.set_parameter("clwstrict", True) # Default None treated as False!
self.assertEqual(str(cmdline).rstrip(), _escape_filename(muscle_exe) +
" -in Fasta/f002 -clwstrict")
self.assertEqual(str(eval(repr(cmdline))), str(cmdline))
child = subprocess.Popen(str(cmdline),
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
universal_newlines=True,
shell=(sys.platform!="win32"))
#Didn't use -quiet so there should be progress reports on stderr,
align = AlignIO.read(child.stdout, "clustal")
align.sort()
self.assertTrue(child.stderr.read().strip().startswith("MUSCLE"))
self.assertEqual(len(records),len(align))
for old, new in zip(records, align):
self.assertEqual(old.id, new.id)
self.assertEqual(str(new.seq).replace("-",""), str(old.seq))
return_code = child.wait()
self.assertEqual(return_code, 0)
child.stdout.close()
child.stderr.close()
del child
def test_simple_clustal_strict(self):
"""Simple muscle call using strict Clustal output"""
input_file = "Fasta/f002"
self.assertTrue(os.path.isfile(input_file))
records = list(SeqIO.parse(input_file, "fasta"))
records.sort(key=lambda rec: rec.id)
#Prepare the command...
cmdline = MuscleCommandline(muscle_exe)
cmdline.set_parameter("in", input_file)
#Use clustal output (with a CLUSTAL header)
cmdline.set_parameter("clwstrict",
True) # Default None treated as False!
self.assertEqual(
str(cmdline).rstrip(),
_escape_filename(muscle_exe) + " -in Fasta/f002 -clwstrict")
self.assertEqual(str(eval(repr(cmdline))), str(cmdline))
child = subprocess.Popen(str(cmdline),
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
universal_newlines=True,
shell=(sys.platform != "win32"))
#Didn't use -quiet so there should be progress reports on stderr,
align = AlignIO.read(child.stdout, "clustal")
align.sort()
self.assertTrue(child.stderr.read().strip().startswith("MUSCLE"))
self.assertEqual(len(records), len(align))
for old, new in zip(records, align):
self.assertEqual(old.id, new.id)
self.assertEqual(str(new.seq).replace("-", ""), str(old.seq))
return_code = child.wait()
self.assertEqual(return_code, 0)
child.stdout.close()
child.stderr.close()
del child
def test_long(self):
"""Simple muscle call using long file."""
# Create a large input file by converting some of another example file
temp_large_fasta_file = "temp_cw_prot.fasta"
records = list(SeqIO.parse("NBRF/Cw_prot.pir", "pir"))[:40]
SeqIO.write(records, temp_large_fasta_file, "fasta")
# Prepare the command...
cmdline = MuscleCommandline(muscle_exe)
cmdline.set_parameter("in", temp_large_fasta_file)
# Use fast options
cmdline.set_parameter("maxiters", 1)
cmdline.set_parameter("diags", True) # Default None treated as False!
# Use clustal output
cmdline.set_parameter("clwstrict",
True) # Default None treated as False!
# Shoudn't need this, but just to make sure it is accepted
cmdline.set_parameter("maxhours", 0.1)
# No progress reports to stderr
cmdline.set_parameter("quiet", True) # Default None treated as False!
self.assertEqual(
str(cmdline).rstrip(),
_escape_filename(muscle_exe) +
" -in temp_cw_prot.fasta -diags -maxhours 0.1" +
" -maxiters 1 -clwstrict -quiet",
)
self.assertEqual(str(eval(repr(cmdline))), str(cmdline))
child = subprocess.Popen(
str(cmdline),
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
universal_newlines=True,
shell=(sys.platform != "win32"),
)
align = AlignIO.read(child.stdout, "clustal")
align.sort()
records.sort(key=lambda rec: rec.id) # noqa: E731
self.assertEqual(len(records), len(align))
for old, new in zip(records, align):
self.assertEqual(old.id, new.id)
self.assertEqual(str(new.seq).replace("-", ""), str(old.seq))
# See if quiet worked:
self.assertEqual("", child.stderr.read().strip())
return_code = child.wait()
self.assertEqual(return_code, 0)
child.stdout.close()
child.stderr.close()
del child
os.remove(temp_large_fasta_file)
def test_Muscle_with_options(self):
"""Round-trip through app with a switch and valued option"""
cmdline = MuscleCommandline(muscle_exe)
cmdline.set_parameter("input", self.infile1) # "input" is alias for "in"
cmdline.set_parameter("out", self.outfile2)
#Use property:
cmdline.objscore = "sp"
cmdline.noanchors = True
self.assertEqual(str(cmdline), _escape_filename(muscle_exe) +
" -in Fasta/f002" +
" -out Fasta/temp_align_out2.fa" +
" -objscore sp -noanchors")
self.assertEqual(str(eval(repr(cmdline))), str(cmdline))
output, error = cmdline()
self.assertEqual(output, "")
self.assertTrue("ERROR" not in error)
self.assertTrue(error.strip().startswith("MUSCLE"), output)
def test_Muscle_with_options(self):
"""Round-trip through app with a switch and valued option"""
cmdline = MuscleCommandline(muscle_exe)
cmdline.set_parameter("input",
self.infile1) #"input" is alias for "in"
cmdline.set_parameter("out", self.outfile2)
#Use property:
cmdline.objscore = "sp"
cmdline.noanchors = True
self.assertEqual(
str(cmdline), muscle_exe + " -in Fasta/f002" +
" -out Fasta/temp_align_out2.fa" + " -objscore sp -noanchors")
self.assertEqual(str(eval(repr(cmdline))), str(cmdline))
output, error = cmdline()
self.assertEqual(output, "")
self.assertTrue("ERROR" not in error)
self.assertTrue(error.strip().startswith("MUSCLE"), output)
def test_Muscle_with_options(self):
"""Round-trip through app with a switch and valued option."""
cmdline = MuscleCommandline(muscle_exe)
cmdline.set_parameter("input", self.infile1) #"input" is alias for "in"
cmdline.set_parameter("out", self.outfile2)
#Use property:
cmdline.objscore = "sp"
cmdline.noanchors = True
self.assertEqual(str(cmdline), muscle_exe +\
" -in Fasta/f002" + \
" -out Fasta/temp_align_out2.fa" + \
" -objscore sp -noanchors")
self.assertEqual(str(eval(repr(cmdline))), str(cmdline))
stdin, stdout, stderr = generic_run(cmdline)
self.assertEqual(stdin.return_code, 0)
self.assertEqual(stdout.read(), "")
self.assert_("ERROR" not in stderr.read())
self.assertEqual(str(stdin._cl), str(cmdline))
def test_long(self):
"""Simple muscle call using long file"""
#Create a large input file by converting some of another example file
temp_large_fasta_file = "temp_cw_prot.fasta"
records = list(SeqIO.parse("NBRF/Cw_prot.pir", "pir"))[:40]
SeqIO.write(records, temp_large_fasta_file, "fasta")
#Prepare the command...
cmdline = MuscleCommandline(muscle_exe)
cmdline.set_parameter("in", temp_large_fasta_file)
#Use fast options
cmdline.set_parameter("maxiters", 1)
cmdline.set_parameter("diags", True) # Default None treated as False!
#Use clustal output
cmdline.set_parameter("clwstrict", True) # Default None treated as False!
#Shoudn't need this, but just to make sure it is accepted
cmdline.set_parameter("maxhours", 0.1)
#No progress reports to stderr
cmdline.set_parameter("quiet", True) # Default None treated as False!
self.assertEqual(str(cmdline).rstrip(), _escape_filename(muscle_exe) +
" -in temp_cw_prot.fasta -diags -maxhours 0.1" +
" -maxiters 1 -clwstrict -quiet")
self.assertEqual(str(eval(repr(cmdline))), str(cmdline))
child = subprocess.Popen(str(cmdline),
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
universal_newlines=True,
shell=(sys.platform!="win32"))
align = AlignIO.read(child.stdout, "clustal")
align.sort()
records.sort(key = lambda rec: rec.id)
self.assertEqual(len(records), len(align))
for old, new in zip(records, align):
self.assertEqual(old.id, new.id)
self.assertEqual(str(new.seq).replace("-",""), str(old.seq))
#See if quiet worked:
self.assertEqual("", child.stderr.read().strip())
return_code = child.wait()
self.assertEqual(return_code, 0)
child.stdout.close()
child.stderr.close()
del child
os.remove(temp_large_fasta_file)
def test_Muscle_with_options(self):
"""Round-trip through app with a switch and valued option"""
cmdline = MuscleCommandline(muscle_exe)
cmdline.set_parameter("input", self.infile1) #"input" is alias for "in"
cmdline.set_parameter("out", self.outfile2)
#Use property:
cmdline.objscore = "sp"
cmdline.noanchors = True
self.assertEqual(str(cmdline), muscle_exe +\
" -in Fasta/f002" + \
" -out Fasta/temp_align_out2.fa" + \
" -objscore sp -noanchors")
self.assertEqual(str(eval(repr(cmdline))), str(cmdline))
child = subprocess.Popen(str(cmdline),
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
shell=(sys.platform!="win32"))
output, error = child.communicate()
self.assertEqual(child.returncode, 0)
self.assertEqual(output, "")
self.assert_("ERROR" not in error)
del child
def test_Muscle_profile_simple(self):
"""Simple round-trip through app doing a profile alignment"""
cmdline = MuscleCommandline(muscle_exe)
cmdline.set_parameter("out", self.outfile3)
cmdline.set_parameter("profile", True)
cmdline.set_parameter("in1", self.infile2)
cmdline.set_parameter("in2", self.infile3)
self.assertEqual(str(cmdline), _escape_filename(muscle_exe) +
" -out Fasta/temp_align_out3.fa" +
" -profile -in1 Fasta/fa01 -in2 Fasta/f001")
self.assertEqual(str(eval(repr(cmdline))), str(cmdline))
output, error = cmdline()
self.assertEqual(output, "")
self.assertTrue("ERROR" not in error)
self.assertTrue(error.strip().startswith("MUSCLE"), output)
def test_Muscle_profile_simple(self):
"""Simple round-trip through app doing a profile alignment"""
cmdline = MuscleCommandline(muscle_exe)
cmdline.set_parameter("out", self.outfile3)
cmdline.set_parameter("profile", True)
cmdline.set_parameter("in1", self.infile2)
cmdline.set_parameter("in2", self.infile3)
self.assertEqual(
str(cmdline), muscle_exe + " -out Fasta/temp_align_out3.fa" +
" -profile -in1 Fasta/fa01 -in2 Fasta/f001")
self.assertEqual(str(eval(repr(cmdline))), str(cmdline))
output, error = cmdline()
self.assertEqual(output, "")
self.assertTrue("ERROR" not in error)
self.assertTrue(error.strip().startswith("MUSCLE"), output)
def test_Muscle_profile_simple(self):
"""Simple round-trip through app doing a profile alignment."""
cmdline = MuscleCommandline(muscle_exe)
cmdline.set_parameter("out", self.outfile3)
cmdline.set_parameter("profile", True)
cmdline.set_parameter("in1", self.infile2)
cmdline.set_parameter("in2", self.infile3)
self.assertEqual(str(cmdline), muscle_exe + \
" -out Fasta/temp_align_out3.fa" + \
" -profile -in1 Fasta/fa01 -in2 Fasta/f001")
self.assertEqual(str(eval(repr(cmdline))), str(cmdline))
stdin, stdout, stderr = generic_run(cmdline)
self.assertEqual(stdin.return_code, 0)
self.assertEqual(stdout.read(), "")
self.assert_("ERROR" not in stderr.read())
self.assertEqual(str(stdin._cl), str(cmdline))
def test_Muscle_profile_simple(self):
"""Simple round-trip through app doing a profile alignment"""
cmdline = MuscleCommandline(muscle_exe)
cmdline.set_parameter("out", self.outfile3)
cmdline.set_parameter("profile", True)
cmdline.set_parameter("in1", self.infile2)
cmdline.set_parameter("in2", self.infile3)
self.assertEqual(str(cmdline), muscle_exe + \
" -out Fasta/temp_align_out3.fa" + \
" -profile -in1 Fasta/fa01 -in2 Fasta/f001")
self.assertEqual(str(eval(repr(cmdline))), str(cmdline))
child = subprocess.Popen(str(cmdline),
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
shell=(sys.platform!="win32"))
output, error = child.communicate()
self.assertEqual(child.returncode, 0)
self.assertEqual(output, "")
self.assert_("ERROR" not in error)
del child
def test_simple_clustal_strict(self):
"""Simple muscle call using strict Clustal output."""
input_file = "Fasta/f002"
self.assert_(os.path.isfile(input_file))
records = list(SeqIO.parse(open(input_file),"fasta"))
#Prepare the command...
cmdline = MuscleCommandline(muscle_exe)
cmdline.set_parameter("in", input_file)
#Preserve input record order (makes checking output easier)
cmdline.set_parameter("stable", True) #Default None treated as False!
#Use clustal output (with a CLUSTAL header)
cmdline.set_parameter("clwstrict", True) #Default None treated as False!
self.assertEqual(str(cmdline).rstrip(), muscle_exe + \
" -in Fasta/f002 -clwstrict -stable")
self.assertEqual(str(eval(repr(cmdline))), str(cmdline))
result, out_handle, err_handle = generic_run(cmdline)
align = AlignIO.read(out_handle, "clustal")
self.assertEqual(len(records),len(align))
for old, new in zip(records, align):
self.assertEqual(old.id, new.id)
self.assertEqual(str(new.seq).replace("-",""), str(old.seq))
#Didn't use -quiet so there should be progress reports on stderr,
self.assert_(err_handle.read().strip().startswith("MUSCLE"))
def test_long(self) :
"""Simple muscle call using long file."""
#Create a large input file by converting some of another example file
temp_large_fasta_file = "temp_cw_prot.fasta"
handle = open(temp_large_fasta_file, "w")
records = list(SeqIO.parse(open("NBRF/Cw_prot.pir", "rU"), "pir"))[:40]
SeqIO.write(records, handle, "fasta")
handle.close()
#Prepare the command...
cmdline = MuscleCommandline(muscle_exe)
cmdline.set_parameter("in", temp_large_fasta_file)
#Preserve input record order
cmdline.set_parameter("stable", True) #Default None treated as False!
#Use fast options
cmdline.set_parameter("maxiters", 1)
cmdline.set_parameter("diags", True) #Default None treated as False!
#Use clustal output
cmdline.set_parameter("clwstrict", True) #Default None treated as False!
#Shoudn't need this, but just to make sure it is accepted
cmdline.set_parameter("maxhours", 0.1)
#No progress reports to stderr
cmdline.set_parameter("quiet", True) #Default None treated as False!
#TODO - Fix the trailing space!
self.assertEqual(str(cmdline).rstrip(), muscle_exe + \
" -in temp_cw_prot.fasta -diags -maxhours 0.1" + \
" -maxiters 1 -clwstrict -stable -quiet")
self.assertEqual(str(eval(repr(cmdline))), str(cmdline))
result, out_handle, err_handle = generic_run(cmdline)
align = AlignIO.read(out_handle, "clustal")
self.assertEqual(len(records), len(align))
for old, new in zip(records, align) :
self.assertEqual(old.id, new.id)
self.assertEqual(str(new.seq).replace("-",""), str(old.seq))
os.remove(temp_large_fasta_file)
#See if quiet worked:
self.assertEqual("", err_handle.read().strip())