def __init__(self, config, prefix, node, log_file=None):
input_file, has_index = self._get_input_file(node)
subnodes, dependencies = [node], node.dependencies
if not has_index:
node = BAMIndexNode(infile=input_file, dependencies=node)
subnodes.append(node)
validation_params = ValidateBAMNode.customize(config=config,
input_bam=input_file,
output_log=log_file,
dependencies=node)
# Check MD tags against reference sequence
# FIXME: Disabled due to issues with Picard/Samtools disagreeing, backwards compatibility.
# validation_params.command.set_kwargs(IN_REFERENCE = prefix["Reference"])
# validation_params.command.add_option("R", "%(IN_REFERENCE)s", sep = "=")
# Ignored since we filter out misses and low-quality hits during mapping, which
# leads to a large proportion of missing mates for PE reads.
validation_params.command.add_option("IGNORE",
"MATE_NOT_FOUND",
sep="=")
# Ignored due to high rate of false positives for lanes with few hits, where
# high-quality reads may case ValidateSamFile to mis-identify the qualities
validation_params.command.add_option("IGNORE",
"INVALID_QUALITY_FORMAT",
sep="=")
subnodes.append(validation_params.build_node())
description = "<w/Validation: " + str(subnodes[0])[1:]
MetaNode.__init__(self,
description=description,
subnodes=subnodes,
dependencies=dependencies)
def __init__(self, config, prefix, node, log_file=None):
input_file, has_index = self._get_input_file(node)
subnodes, dependencies = [node], node.dependencies
if not has_index:
node = BAMIndexNode(infile=input_file,
dependencies=node)
subnodes.append(node)
validation_params = ValidateBAMNode.customize(config=config,
input_bam=input_file,
output_log=log_file,
dependencies=node)
# Check MD tags against reference sequence
# FIXME: Disabled due to issues with Picard/Samtools disagreeing,
# backwards compatibility. See the discussion at
# http://sourceforge.net/mailarchive/message.php?msg_id=31348639
# validation_params.command.set_kwargs(IN_REF=prefix["Reference"])
# validation_params.command.add_option("R", "%(IN_REF)s", sep="=")
# Ignored since we may filter out misses and low-quality hits during
# mapping, which leads to a large proportion of missing PE mates.
validation_params.command.add_option("IGNORE", "MATE_NOT_FOUND",
sep="=")
# Ignored due to high rate of false positives for lanes with few hits,
# where high-quality reads may cause mis-identification of qualities
validation_params.command.add_option("IGNORE",
"INVALID_QUALITY_FORMAT", sep="=")
subnodes.append(validation_params.build_node())
description = "<w/Validation: " + str(subnodes[0])[1:]
MetaNode.__init__(self,
description=description,
subnodes=subnodes,
dependencies=dependencies)
def __init__(self, config, reference, infiles, outfile, intervals=None,
dependencies=()):
if not intervals:
intervals = outfile + ".intervals"
infiles = safe_coerce_to_tuple(infiles)
trainer = _IndelTrainerNode(config=config,
reference=reference,
infiles=infiles,
outfile=intervals,
dependencies=dependencies)
aligner = _IndelRealignerNode(config=config,
reference=reference,
intervals=intervals,
infiles=infiles,
outfile=outfile,
dependencies=trainer)
description = "<GATK Indel Realigner: %i files -> '%s'>" \
% (len(infiles), outfile)
MetaNode.__init__(self,
description=description,
subnodes=[trainer, aligner],
dependencies=dependencies)
def __init__(self, input_files, destination, filter_by, dependencies=()):
subnodes = []
filter_by = dict(filter_by)
for (filename, node) in input_files.iteritems():
output_filename = fileutils.reroot_path(destination, filename)
subnodes.append(FilterSingletonsNode(input_file=filename,
output_file=output_filename,
filter_by=filter_by,
dependencies=node))
MetaNode.__init__(self,
description="<FilterSingleton: %i files -> '%s'>"
% (len(subnodes), destination),
subnodes=subnodes,
dependencies=dependencies)
def __init__(self, input_files, destination, filter_by, dependencies=()):
subnodes = []
filter_by = dict(filter_by)
for (filename, node) in input_files.iteritems():
output_filename = fileutils.reroot_path(destination, filename)
subnodes.append(FilterSingletonsNode(input_file=filename,
output_file=output_filename,
filter_by=filter_by,
dependencies=node))
MetaNode.__init__(self,
description="<FilterSingleton: %i files -> '%s'>"
% (len(subnodes), destination),
subnodes=subnodes,
dependencies=dependencies)
def __init__(self, rootdir, sequences, preset = "auto", subnodes = (), dependencies = ()):
subnodes = []
for sequence in sequences:
prefix = os.path.join(rootdir, sequence)
node = MAFFTNode(infile = prefix + ".fasta",
outfile = prefix + ".afa",
preset = preset,
dependencies = dependencies)
subnodes.append(node)
MetaNode.__init__(self,
description = "<MAFFTAlignSequences (%s): In '%s'>" \
% (preset, rootdir),
subnodes = subnodes,
dependencies = dependencies)
def __init__(self, config, reference, infiles, outfile, intervals=None, dependencies=()):
if not intervals:
intervals = outfile + ".intervals"
infiles = safe_coerce_to_tuple(infiles)
trainer = _IndelTrainerNode(
config=config, reference=reference, infiles=infiles, outfile=intervals, dependencies=dependencies
)
aligner = _IndelRealignerNode(
config=config,
reference=reference,
intervals=intervals,
infiles=infiles,
outfile=outfile,
dependencies=trainer,
)
description = "<GATK Indel Realigner: %i files -> '%s'>" % (len(infiles), outfile)
MetaNode.__init__(self, description=description, subnodes=[trainer, aligner], dependencies=dependencies)
