def finish(self):
# Once executed, alignment is converted into relaxed
# interleaved phylip format.
alg = SeqGroup(os.path.join(self.jobs[0].jobdir, "alg.fasta"))
fasta = alg.write(format="fasta")
phylip = alg.write(format="iphylip_relaxed")
AlgTask.store_data(self, fasta, phylip)
def finish(self):
# Once executed, alignment is converted into relaxed
# interleaved phylip format.
final_job = self.jobs[2]
alg = SeqGroup(os.path.join(final_job.jobdir, "alg.fasta"))
alg.write(outfile=self.alg_fasta_file, format="fasta")
alg.write(outfile=self.alg_phylip_file, format="iphylip_relaxed")
AlgTask.finish(self)
def finish(self):
# Once executed, alignment is converted into relaxed
# interleaved phylip format.
alg_file = os.path.join(self.jobs[0].jobdir, "alg.fasta")
# ClustalO returns a tricky fasta file
alg = read_fasta(alg_file, header_delimiter=" ")
fasta = alg.write(format="fasta")
phylip = alg.write(format="iphylip_relaxed")
AlgTask.store_data(self, fasta, phylip)
def __init__(self, nodeid, multiseq_file, seqtype, conf, confname):
self.confname = confname
self.conf = conf
# Initialize task
AlgTask.__init__(self, nodeid, "alg", "Meta-Alg",
OrderedDict(), self.conf[self.confname])
self.seqtype = seqtype
self.multiseq_file = multiseq_file
self.size = conf["_nodeinfo"][nodeid].get("size", 0)
self.all_alg_files = None
self.init()
def finish(self):
# Once executed, alignment is converted into relaxed
# interleaved phylip format.
alg = SeqGroup(os.path.join(self.jobs[0].jobdir, "mcoffee.fasta"))
fasta = alg.write(format="fasta")
phylip = alg.write(format="iphylip_relaxed")
alg_list_string = '\n'.join([pjoin(GLOBALS["input_dir"],
aname) for aname in self.all_alg_files])
db.add_task_data(self.taskid, DATATYPES.alg_list, alg_list_string)
AlgTask.store_data(self, fasta, phylip)
def __init__(self, nodeid, multiseq_file, seqtype, conf, confname):
GLOBALS["citator"].add(MAFFT_CITE)
self.confname = confname
self.conf = conf
# Initialize task
AlgTask.__init__(self, nodeid, "alg", "Mafft",
OrderedDict(), self.conf[confname])
self.seqtype = seqtype
self.multiseq_file = multiseq_file
self.init()
def __init__(self, nodeid, multiseq_file, seqtype, conf, confname):
GLOBALS["citator"].add(DIALIGN_CITE)
# fixed options for running this task
base_args = OrderedDict({"": None})
# Initialize task
self.confname = confname
self.conf = conf
AlgTask.__init__(self, nodeid, "alg", "DialignTX", base_args, self.conf[self.confname])
self.seqtype = seqtype
self.multiseq_file = multiseq_file
self.init()
def __init__(self, nodeid, multiseq_file, seqtype, conf):
# Initialize task
AlgTask.__init__(self, nodeid, "alg", "Usearch-Uhire",
OrderedDict(), conf["uhire"])
self.conf = conf
self.seqtype = seqtype
self.multiseq_file = multiseq_file
self.init()
self.alg_fasta_file = os.path.join(self.taskdir, "final_alg.fasta")
self.alg_phylip_file = os.path.join(self.taskdir, "final_alg.iphylip")
def __init__(self, nodeid, seqtype, all_alg_files, conf, confname, parent_ids):
GLOBALS["citator"].add(MCOFFEE_CITE)
base_args = OrderedDict({
"-output": "fasta",
})
# Initialize task
self.confname = confname
self.conf = conf
AlgTask.__init__(self, nodeid, "alg", "Mcoffee",
base_args, self.conf[confname])
self.all_alg_files = all_alg_files
self.parent_ids = parent_ids
self.seqtype = seqtype
self.init()
def finish(self):
if self.conf[self.confname]["_alg_trimming"]:
# If trimming happened after mcoffee, let's save the
# resulting output
trim_job = self.jobs[-1]
alg = SeqGroup(pjoin(trim_job.jobdir, trim_job.alg_fasta_file))
fasta = alg.write(format="fasta")
phylip = alg.write(format="iphylip_relaxed")
AlgTask.store_data(self, fasta, phylip)
else:
# If no post trimming, output is just what Mcoffee
# produced, so we can recycle its data ids.
mc_task = self.jobs[-1]
fasta_id = db.get_dataid(mc_task.taskid, DATATYPES.alg_fasta)
phylip_id = db.get_dataid(mc_task.taskid, DATATYPES.alg_phylip)
db.register_task_data(self.taskid, DATATYPES.alg_fasta, fasta_id)
db.register_task_data(self.taskid, DATATYPES.alg_phylip, phylip_id)
def __init__(self, nodeid, multiseq_file, seqtype, conf, confname):
GLOBALS["citator"].add(MUSCLE_CITE)
# fixed Muscle options
base_args = OrderedDict({
'-in': None,
'-out': None,
})
self.confname = confname
self.conf = conf
# Initialize task
AlgTask.__init__(self, nodeid, "alg", "Muscle",
base_args, self.conf[confname])
self.seqtype = seqtype
self.multiseq_file = multiseq_file
self.init()
def __init__(self, nodeid, multiseq_file, seqtype, conf, confname):
GLOBALS["citator"].add(CLUSTALO_CITE)
base_args = OrderedDict({
'-i': None,
'-o': None,
'--outfmt': "fa",
})
self.confname = confname
self.conf = conf
# Initialize task
AlgTask.__init__(self, nodeid, "alg", "Clustal-Omega",
base_args, self.conf[self.confname])
self.seqtype = seqtype
self.multiseq_file = multiseq_file
self.init()
def init_output_info(self):
self.alg_list_file = "%s.%s" %(self.taskid, DATATYPES.alg_list)
AlgTask.init_output_info(self)
def finish(self):
alg = SeqGroup(os.path.join(self.jobs[0].jobdir, "alg.fasta"))
fasta = alg.write(format="fasta")
phylip = alg.write(format="iphylip_relaxed")
AlgTask.store_data(self, fasta, phylip)