def check_EMBOSS_to_AlignIO(self, filename, old_format, skip_formats=()):
"""Check AlignIO can read seqret's conversion of the file."""
self.assertTrue(os.path.isfile(filename), filename)
old_aligns = list(AlignIO.parse(filename, old_format))
formats = ["clustal", "phylip", "ig", "msf"]
if len(old_aligns) == 1:
formats.extend(["fasta", "nexus"])
for new_format in formats:
if new_format in skip_formats:
continue
cline = SeqretCommandline(
exes["seqret"],
sequence=filename,
sformat=old_format,
osformat=new_format,
auto=True, # no prompting
stdout=True,
)
# Run the tool,
with subprocess.Popen(
str(cline),
stdin=subprocess.PIPE,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
universal_newlines=True,
shell=(sys.platform != "win32"),
) as child:
child.stdin.close()
new_aligns = list(AlignIO.parse(child.stdout, new_format))
msg = "converting %s from %s to %s" % (filename, old_format,
new_format)
self.compare_alignments(old_aligns, new_aligns, msg)
def emboss_piped_SeqIO_convert(records, old_format, new_format):
"""Run seqret, returns records (as a generator)."""
# Setup, this assumes for all the format names used
# Biopython and EMBOSS names are consistent!
cline = SeqretCommandline(
exes["seqret"],
sformat=old_format,
osformat=new_format,
auto=True, # no prompting
filter=True,
)
# Run the tool,
child = subprocess.Popen(
str(cline),
stdin=subprocess.PIPE,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
universal_newlines=True,
shell=(sys.platform != "win32"),
)
SeqIO.write(records, child.stdin, old_format)
child.stdin.close()
child.stderr.close()
# TODO - Is there a nice way to return an iterator AND
# automatically close the handle?
records = list(SeqIO.parse(child.stdout, new_format))
child.stdout.close()
return records
def check_EMBOSS_to_SeqIO(self, filename, old_format, skip_formats=()):
"""Check SeqIO can read read seqret's conversion output."""
# TODO: Why can't we read EMBOSS's swiss output?
self.assertTrue(os.path.isfile(filename))
old_records = list(SeqIO.parse(filename, old_format))
for new_format in ["genbank", "fasta", "pir", "embl", "ig"]:
if new_format in skip_formats:
continue
cline = SeqretCommandline(
exes["seqret"],
sequence=filename,
sformat=old_format,
osformat=new_format,
auto=True, # no prompting
stdout=True,
)
# Run the tool,
with subprocess.Popen(
str(cline),
stdin=subprocess.PIPE,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
universal_newlines=True,
shell=(sys.platform != "win32"),
) as child:
child.stdin.close()
new_records = list(SeqIO.parse(child.stdout, new_format))
msg = "converting %s from %s to %s" % (filename, old_format,
new_format)
self.compare_records(old_records, new_records, msg)
def emboss_piped_SeqIO_convert(records, old_format, new_format):
"""Run seqret, returns records (as a generator)."""
#Setup, this assumes for all the format names used
#Biopython and EMBOSS names are consistent!
cline = SeqretCommandline(
exes["seqret"],
sformat=old_format,
osformat=new_format,
auto=True, #no prompting
filter=True)
#Run the tool,
child = subprocess.Popen(str(cline),
stdin=subprocess.PIPE,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
shell=(sys.platform != "win32"))
SeqIO.write(records, child.stdin, old_format)
child.stdin.close()
return SeqIO.parse(child.stdout, new_format)
def emboss_convert(filename, old_format, new_format):
"""Run seqret, returns handle."""
#Setup, this assumes for all the format names used
#Biopython and EMBOSS names are consistent!
cline = SeqretCommandline(
exes["seqret"],
sequence=filename,
sformat=old_format,
osformat=new_format,
auto=True, #no prompting
stdout=True)
#Run the tool,
child = subprocess.Popen(str(cline),
stdin=subprocess.PIPE,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
shell=(sys.platform != "win32"))
child.stdin.close()
return child.stdout
def emboss_piped_AlignIO_convert(alignments, old_format, new_format):
"""Run seqret, returns alignments (as a generator)."""
#Setup, this assumes for all the format names used
#Biopython and EMBOSS names are consistent!
cline = SeqretCommandline(exes["seqret"],
sformat = old_format,
osformat = new_format,
auto = True, #no prompting
filter = True)
#Run the tool,
child = subprocess.Popen(str(cline),
stdin=subprocess.PIPE,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
universal_newlines=True,
shell=(sys.platform!="win32"))
try:
AlignIO.write(alignments, child.stdin, old_format)
except Exception, err:
child.stdin.close()
child.stderr.close()
child.stdout.close()
raise
def test_abi(self):
"""Check SeqIO agrees with EMBOSS' Abi to FASTQ conversion."""
# This lets use check the id, sequence, and quality scores
for filename in ["Abi/3730.ab1", "Abi/empty.ab1"]:
old = SeqIO.read(filename, "abi")
cline = SeqretCommandline(
exes["seqret"],
sequence=filename,
sformat="abi",
osformat="fastq-sanger",
auto=True, # no prompting
stdout=True,
)
# Run the tool,
with subprocess.Popen(
str(cline),
stdin=subprocess.PIPE,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
universal_newlines=True,
shell=(sys.platform != "win32"),
) as child:
child.stdin.close()
new = SeqIO.read(child.stdout, "fastq-sanger")
if emboss_version == (6, 4, 0) and new.id == "EMBOSS_001":
# Avoid bug in EMBOSS 6.4.0 (patch forthcoming)
pass
else:
self.assertEqual(old.id, new.id)
self.assertEqual(str(old.seq), str(new.seq))
if emboss_version < (6, 3, 0) and new.letter_annotations[
"phred_quality"] == [1] * len(old):
# Apparent bug in EMBOSS 6.2.0.1 on Windows
pass
else:
self.assertEqual(old.letter_annotations,
new.letter_annotations)
