def _determine_lane_type(prefixes, data, path):
if isinstance(data, types.StringTypes):
return "Raw"
elif isinstance(data, types.DictType):
if all((key in _READ_TYPES) for key in data):
for (key, files) in data.iteritems():
is_paired = paths.is_paired_end(files)
if is_paired and (key != "Paired"):
raise MakefileError("Error at Barcode level; Path "
"includes {Pair} key, but read-type "
"is not Paired:\n "
"%s:%s" % (":".join(path), key))
elif not is_paired and (key == "Paired"):
raise MakefileError("Error at Barcode level; Paired pre-"
"trimmed reads specified, but path "
"does not contain {Pair} key:\n "
"%s:%s" % (":".join(path), key))
return "Trimmed"
elif all((key in prefixes) for key in data):
return "BAMs"
raise MakefileError("Error at Barcode level; keys must either be "
"prefix-names, OR 'Paired', 'Single' or 'Collapsed'. "
"Found: %s" % (", ".join(data),))
def _build_bwa_nodes(config, parameters, input_filename, tags, options):
if paths.is_paired_end(input_filename):
input_file_1 = input_file_2 = input_filename
aln_keys, sam_key = ("aln_1", "aln_2"), "sampe"
else:
input_file_1, input_file_2 = input_filename, ""
aln_keys, sam_key = ("aln",), "samse"
params = BWANode.customize(input_file_1 = input_file_1.format(Pair = 1),
input_file_2 = input_file_2.format(Pair = 2),
threads = config.bwa_max_threads,
**parameters)
for aln_key in aln_keys:
if not options["Aligners"]["BWA"]["UseSeed"]:
params.commands[aln_key].set_option("-l", 2**16 - 1)
if options["QualityOffset"] in (64, "Solexa"):
params.commands[aln_key].set_option("-I")
_apply_aln_user_parameters(options["Aligners"]["BWA"], params, aln_keys)
read_group = "@RG\tID:{ID}\tSM:{SM}\tLB:{LB}\tPU:{PU_src}\tPL:{PL}\tPG:{PG_lc}".format(PG_lc = tags["PG"].lower(), **tags)
params.commands[sam_key].set_option("-r", read_group)
params.commands["convert"].set_option('-q', options["Aligners"]["BWA"]["MinQuality"])
cl_tag = _build_bwa_cl_tag(options, sam_key)
pg_tags = ("bwa:CL:%s" % (cl_tag.replace("%", "%%"),))
params.commands["convert"].add_option("--update-pg-tag", pg_tags)
return params.build_node()
def _build_bowtie2_nodes(config, parameters, input_filename, tags, options):
if paths.is_paired_end(input_filename):
input_filename_1 = input_filename_2 = input_filename
else:
input_filename_1, input_filename_2 = input_filename, ""
params = Bowtie2Node.customize(input_file_1 = input_filename_1.format(Pair = 1),
input_file_2 = input_filename_2.format(Pair = 2),
threads = config.bowtie2_max_threads,
**parameters)
params.commands["convert"].set_option('-q', options["Aligners"]["Bowtie2"]["MinQuality"])
if options["QualityOffset"] == 64:
params.commands["aln"].set_option("--phred64")
elif options["QualityOffset"] == 33:
params.commands["aln"].set_option("--phred33")
else:
params.commands["aln"].set_option("--solexa-quals")
_apply_aln_user_parameters(options["Aligners"]["Bowtie2"], params, ("aln",))
pg_tag = "bowtie2:CL:%s" % (_build_bowtie_cl_tag(options).replace("%", "%%"),)
params.commands["convert"].add_option("--update-pg-tag", pg_tag)
params.commands["aln"].set_option("--rg-id", tags["ID"])
for tag_name in ("SM", "LB", "PU", "PL", "PG"):
if tag_name == "PG":
tag_value = tags["PG"].lower()
elif tag_name == "PU":
tag_value = tags["PU_src"]
else:
tag_value = tags[tag_name]
params.commands["aln"].add_option("--rg", "%s:%s" % (tag_name, tag_value))
return params.build_node()
def _determine_lane_type(prefixes, data, path):
if isinstance(data, types.StringTypes):
return "Raw"
elif isinstance(data, types.DictType):
if all((key in _READ_TYPES) for key in data):
for (key, files) in data.iteritems():
is_paired = paths.is_paired_end(files)
if is_paired and (key != "Paired"):
raise MakefileError("Error at Barcode level; Path "
"includes {Pair} key, but read-type "
"is not Paired:\n "
"%s:%s" % (":".join(path), key))
elif not is_paired and (key == "Paired"):
raise MakefileError("Error at Barcode level; Paired pre-"
"trimmed reads specified, but path "
"does not contain {Pair} key:\n "
"%s:%s" % (":".join(path), key))
return "Trimmed"
elif all((key in prefixes) for key in data):
return "BAMs"
raise MakefileError("Error at Barcode level; keys must either be "
"prefix-names, OR 'Paired', 'Single' or 'Collapsed'. "
"Found: %s" % (", ".join(data), ))
def _build_bwa_nodes(config, parameters, input_filename, tags, options):
if paths.is_paired_end(input_filename):
input_file_1 = input_file_2 = input_filename
aln_keys, sam_key = ("aln_1", "aln_2"), "sampe"
else:
input_file_1, input_file_2 = input_filename, ""
aln_keys, sam_key = ("aln", ), "samse"
params = BWANode.customize(input_file_1=input_file_1.format(Pair=1),
input_file_2=input_file_2.format(Pair=2),
threads=config.bwa_max_threads,
**parameters)
for aln_key in aln_keys:
if not options["Aligners"]["BWA"]["UseSeed"]:
params.commands[aln_key].set_option("-l", 2**16 - 1)
if options["QualityOffset"] in (64, "Solexa"):
params.commands[aln_key].set_option("-I")
_apply_aln_user_parameters(options["Aligners"]["BWA"], params, aln_keys)
read_group = "@RG\tID:{ID}\tSM:{SM}\tLB:{LB}\tPU:{PU_src}\tPL:{PL}\tPG:{PG_lc}".format(
PG_lc=tags["PG"].lower(), **tags)
params.commands[sam_key].set_option("-r", read_group)
params.commands["convert"].set_option(
'-q', options["Aligners"]["BWA"]["MinQuality"])
cl_tag = _build_bwa_cl_tag(options, sam_key)
pg_tags = ("bwa:CL:%s" % (cl_tag.replace("%", "%%"), ))
params.commands["convert"].add_option("--update-pg-tag", pg_tags)
return params.build_node()
def _init_unaligned_lane(self, config, prefix, record):
aln_key, aln_func = _select_aligner(record["Options"])
prefix_key = "Node:%s" % (aln_key, )
postfix = [
"minQ%i" % record["Options"]["Aligners"][aln_key]["MinQuality"]
]
if not record["Options"]["Aligners"][aln_key].get("UseSeed", True):
postfix.append("noSeed")
for (key, input_filename) in self.reads.files.iteritems():
# Common parameters between BWA / Bowtie2
output_filename = os.path.join(
self.folder, "%s.%s.bam" % (key.lower(), ".".join(postfix)))
parameters = {
"output_file": output_filename,
"prefix": prefix["Path"],
"reference": prefix["Reference"],
"input_file_1": input_filename,
"input_file_2": None,
"dependencies": self.reads.nodes + (prefix[prefix_key], )
}
if paths.is_paired_end(input_filename):
parameters["input_file_1"] = input_filename.format(Pair=1)
parameters["input_file_2"] = input_filename.format(Pair=2)
alignment_obj = aln_func(config=config,
parameters=parameters,
tags=self.tags,
options=record["Options"])
alignment_opts = record["Options"]["Aligners"][aln_key]
alignment_obj.commands["convert"].set_option(
'-q', alignment_opts["MinQuality"])
if alignment_opts["FilterUnmappedReads"]:
alignment_obj.commands["convert"].set_option('-F', "0x4")
alignment_node = alignment_obj.build_node()
validated_node = index_and_validate_bam(config, prefix,
alignment_node)
self.bams[key] = {output_filename: validated_node}
def _build_bowtie2_nodes(config, parameters, input_filename, tags, options):
if paths.is_paired_end(input_filename):
input_filename_1 = input_filename_2 = input_filename
else:
input_filename_1, input_filename_2 = input_filename, ""
params = Bowtie2Node.customize(
input_file_1=input_filename_1.format(Pair=1),
input_file_2=input_filename_2.format(Pair=2),
threads=config.bowtie2_max_threads,
**parameters)
params.commands["convert"].set_option(
'-q', options["Aligners"]["Bowtie2"]["MinQuality"])
if options["QualityOffset"] == 64:
params.commands["aln"].set_option("--phred64")
elif options["QualityOffset"] == 33:
params.commands["aln"].set_option("--phred33")
else:
params.commands["aln"].set_option("--solexa-quals")
_apply_aln_user_parameters(options["Aligners"]["Bowtie2"], params,
("aln", ))
pg_tag = "bowtie2:CL:%s" % (_build_bowtie_cl_tag(options).replace(
"%", "%%"), )
params.commands["convert"].add_option("--update-pg-tag", pg_tag)
params.commands["aln"].set_option("--rg-id", tags["ID"])
for tag_name in ("SM", "LB", "PU", "PL", "PG"):
if tag_name == "PG":
tag_value = tags["PG"].lower()
elif tag_name == "PU":
tag_value = tags["PU_src"]
else:
tag_value = tags[tag_name]
params.commands["aln"].add_option("--rg",
"%s:%s" % (tag_name, tag_value))
return params.build_node()
def _init_unaligned_lane(self, config, prefix, record):
aln_key, aln_func = _select_aligner(record["Options"])
prefix_key = "Node:%s" % (aln_key,)
postfix = ["minQ%i" % record["Options"]["Aligners"][aln_key]["MinQuality"]]
if not record["Options"]["Aligners"][aln_key].get("UseSeed", True):
postfix.append("noSeed")
for (key, input_filename) in self.reads.files.iteritems():
# Common parameters between BWA / Bowtie2
output_filename = os.path.join(self.folder, "%s.%s.bam" % (key.lower(), ".".join(postfix)))
parameters = {"output_file" : output_filename,
"prefix" : prefix["Path"],
"reference" : prefix["Reference"],
"input_file_1" : input_filename,
"input_file_2" : None,
"dependencies" : self.reads.nodes + (prefix[prefix_key],)}
if paths.is_paired_end(input_filename):
parameters["input_file_1"] = input_filename.format(Pair = 1)
parameters["input_file_2"] = input_filename.format(Pair = 2)
alignment_obj = aln_func(config = config,
parameters = parameters,
tags = self.tags,
options = record["Options"])
alignment_opts = record["Options"]["Aligners"][aln_key]
alignment_obj.commands["convert"].set_option('-q', alignment_opts["MinQuality"])
if alignment_opts["FilterUnmappedReads"]:
alignment_obj.commands["convert"].set_option('-F', "0x4")
alignment_node = alignment_obj.build_node()
validated_node = index_and_validate_bam(config, prefix, alignment_node)
self.bams[key] = {output_filename : validated_node}
