def _get_common_parameters(version):
global _DEPRECATION_WARNING_PRINTED
if version == VERSION_14:
version_check = _VERSION_14_CHECK
elif version == VERSION_15:
version_check = _VERSION_15_CHECK
else:
raise CmdError("Unknown version: %s" % version)
cmd = AtomicCmdBuilder("AdapterRemoval", CHECK_VERSION=version_check)
# Trim Ns at read ends
cmd.set_option("--trimns", fixed=False)
# Trim low quality scores
cmd.set_option("--trimqualities", fixed=False)
try:
if not _DEPRECATION_WARNING_PRINTED and version_check.version < (2, 0):
import pypeline.ui as ui
ui.print_warn("\nWARNING: AdapterRemoval v1.5.x is deprecated;")
ui.print_warn(" Upgrading to 2.1.x is strongly adviced!\n")
ui.print_warn(
" Download the newest version of AdapterRemoval at ")
ui.print_warn(
" https://github.com/MikkelSchubert/adapterremoval\n")
_DEPRECATION_WARNING_PRINTED = True
except versions.VersionRequirementError:
pass
return cmd
def __init__(self, parameters):
self._version = parameters.version
self._basename = parameters.basename
if len(parameters.input_files_1) != len(parameters.input_files_2):
raise CmdError("Number of mate 1 files differ from mate 2 files: %i != %i" \
% (len(parameters.input_files_1),
len(parameters.input_files_2)))
zcat_pair_1 = _build_unicat_command(parameters.input_files_1,
"uncompressed_input_1")
zcat_pair_2 = _build_unicat_command(parameters.input_files_2,
"uncompressed_input_2")
zip_pair_1 = _build_zip_command(parameters.output_format,
parameters.output_prefix,
".pair1.truncated")
zip_pair_2 = _build_zip_command(parameters.output_format,
parameters.output_prefix,
".pair2.truncated")
zip_discarded = _build_zip_command(parameters.output_format,
parameters.output_prefix,
".discarded")
adapterrm = parameters.command.finalize()
commands = [adapterrm, zip_pair_1, zip_pair_2]
if parameters.version == VERSION_15:
zip_aln = _build_zip_command(parameters.output_format,
parameters.output_prefix,
".collapsed")
zip_aln_trunc = _build_zip_command(parameters.output_format,
parameters.output_prefix,
".collapsed.truncated")
zip_unaligned = _build_zip_command(parameters.output_format,
parameters.output_prefix,
".singleton.truncated")
commands += [zip_aln, zip_aln_trunc, zip_unaligned]
else:
zip_aln = _build_zip_command(parameters.output_format,
parameters.output_prefix,
".singleton.aln.truncated")
zip_unaligned = _build_zip_command(parameters.output_format,
parameters.output_prefix,
".singleton.unaln.truncated")
commands += [zip_aln, zip_unaligned]
commands += [zip_discarded, zcat_pair_1, zcat_pair_2]
# Opening of pipes block, so the order of these commands is dependent upon
# the order of file-opens in atomiccmd and the the programs themselves.
commands = ParallelCmds(commands)
description = "<PE_AdapterRM: %s -> '%s.*'>" \
% (fileutils.describe_files(parameters.input_files_1).replace("file", "pair"),
parameters.output_prefix)
CommandNode.__init__(self,
command=commands,
description=description,
dependencies=parameters.dependencies)
def __init__(self, commands):
self._ready = False
commands = safe_coerce_to_tuple(commands)
for command in commands:
if not isinstance(command, (AtomicCmd, _CommandSet)):
raise CmdError(
"ParallelCmds must only contain AtomicCmds or other ParallelCmds!"
)
_CommandSet.__init__(self, commands)
def _are_fastq_checks_required(version):
if version == VERSION_14:
return True
elif version == VERSION_15:
try:
return _VERSION_15_CHECK.version < (2, 0)
except versions.VersionRequirementError:
return True
else:
raise CmdError("Unknown version: %s" % version)
def _build_zip_command(output_format, prefix, name, output=None):
if output_format == "bz2":
command, ext = "bzip2", ".bz2"
elif output_format == "gz":
command, ext = "gzip", ".gz"
else:
raise CmdError("Invalid output-format (%s), please select 'gz' or 'bz2'" \
% repr(output_format))
basename = os.path.basename(prefix)
return AtomicCmd([command, "-c"],
TEMP_IN_STDIN=basename + name,
OUT_STDOUT=prefix + (output or name) + ext)
def _build_zip_command(output_format, prefix, name, output=None):
if output_format not in ("gz", "bz2"):
message = "Invalid output-format (%r), please select 'gz' or 'bz2'"
raise CmdError(message % (output_format, ))
basename = os.path.basename(prefix)
compress = factory.new("zip")
compress.set_option("--format", output_format)
compress.add_value("%(TEMP_IN_PIPE)s")
compress.set_kwargs(TEMP_IN_PIPE=basename + name,
OUT_STDOUT=prefix + (output or name) + "." +
output_format)
return compress.finalize()
def _get_common_parameters(version):
if version == VERSION_14:
version_check = _VERSION_14_CHECK
elif version == VERSION_15:
version_check = _VERSION_15_CHECK
else:
raise CmdError("Unknown version: %s" % version)
cmd = AtomicCmdBuilder("AdapterRemoval", CHECK_VERSION=version_check)
# Trim Ns at read ends
cmd.set_option("--trimns", fixed=False)
# Trim low quality scores
cmd.set_option("--trimqualities", fixed=False)
return cmd
def _validate_commands(self):
if len(self._commands) != len(set(self._commands)):
raise ValueError("Same command included multiple times in %s" \
% (self.__class__.__name__,))
filenames = collections.defaultdict(int)
for command in self._commands:
for filename in command.expected_temp_files:
filenames[filename] += 1
for filename in command.optional_temp_files:
filenames[filename] += 1
clobbered = [
filename for (filename, count) in filenames.items() if (count > 1)
]
if any(clobbered):
raise CmdError("Commands clobber each others' files: %s" %
(", ".join(clobbered), ))
def _get_common_parameters(version):
if version == VERSION_14:
version_check = _VERSION_14_CHECK
elif version == VERSION_15:
version_check = _VERSION_15_CHECK
else:
raise CmdError("Unknown version: %s" % version)
cmd = AtomicCmdBuilder("AdapterRemoval", CHECK_VERSION=version_check)
# Allow 1/3 mismatches in the aligned region
cmd.set_option("--mm", 3, fixed=False)
# Minimum length of trimmed reads
cmd.set_option("--minlength", 25, fixed=False)
# Trim Ns at read ends
cmd.set_option("--trimns", fixed=False)
# Trim low quality scores
cmd.set_option("--trimqualities", fixed=False)
# Offset of quality scores
cmd.set_option("--qualitybase", 33, fixed=False)
return cmd
def stdout(self):
raise CmdError("%s does not implement property 'stdout'!" \
% (self.__class__.__name__,))
def __init__(self, commands):
self._commands = safe_coerce_to_tuple(commands)
if not self._commands:
raise CmdError("Empty list passed to command set")
self._validate_commands()
def __init__(self, parameters):
self._quality_offset = parameters.quality_offset
self._version = parameters.version
self._basename = parameters.basename
self._collapse = parameters.collapse
self._check_fastqs = _are_fastq_checks_required(parameters.version)
if len(parameters.input_files_1) != len(parameters.input_files_2):
raise CmdError(
"Number of mate 1 files differ from mate 2 files: "
"%i != %i" %
(len(parameters.input_files_1), len(parameters.input_files_2)))
zcat_pair_1 = _build_cat_command(parameters.input_files_1,
"uncompressed_input_1")
zcat_pair_2 = _build_cat_command(parameters.input_files_2,
"uncompressed_input_2")
zip_pair_1 = _build_zip_command(parameters.output_format,
parameters.output_prefix,
".pair1.truncated")
zip_pair_2 = _build_zip_command(parameters.output_format,
parameters.output_prefix,
".pair2.truncated")
zip_discarded = _build_zip_command(parameters.output_format,
parameters.output_prefix,
".discarded")
adapterrm = parameters.command.finalize()
commands = [adapterrm, zip_pair_1, zip_pair_2]
if parameters.version == VERSION_15:
zip_unaligned = _build_zip_command(parameters.output_format,
parameters.output_prefix,
".singleton.truncated")
if parameters.collapse:
zip_aln = _build_zip_command(parameters.output_format,
parameters.output_prefix,
".collapsed")
zip_aln_trunc = _build_zip_command(parameters.output_format,
parameters.output_prefix,
".collapsed.truncated")
commands += [zip_aln, zip_aln_trunc, zip_unaligned]
else:
commands += [zip_unaligned]
else:
zip_aln = _build_zip_command(parameters.output_format,
parameters.output_prefix,
".singleton.aln.truncated")
zip_unaligned = _build_zip_command(parameters.output_format,
parameters.output_prefix,
".singleton.unaln.truncated")
commands += [zip_aln, zip_unaligned]
commands += [zip_discarded, zcat_pair_1, zcat_pair_2]
commands = ParallelCmds(commands)
description = "<AdapterRM (PE): %s -> '%s.*'>" \
% (fileutils.describe_paired_files(parameters.input_files_1,
parameters.input_files_2),
parameters.output_prefix)
CommandNode.__init__(self,
command=commands,
description=description,
dependencies=parameters.dependencies)