def nonltr(self):
print 'nonltr: starting'
start = time.time()
# nonltr
#cmd0 = self.base_path + "/nonltr/run_MGEScan.pl \
# -genome=%(genome_dir)s \
# -data=%(data_dir)s \
# -hmmerv=%(hmmerv)s"
cmd0 = "python " + self.base_path + "/nonltr/nonltr.py " + \
"%(genome_dir)s " + \
"%(data_dir)s "
if self.mpi_enabled:
#cmd0 = (cmd0 + " -mpi=%(mpi_enabled)s")
cmd0 = (cmd0 + " --mpi=%(mpi_enabled)s")
res0 = self.run_cmd(cmd0)
# gff3
self.nonltr_out_path = utils.get_abspath(self.data_dir + "/info/full/")
self.nonltr_gff_path = utils.get_abspath(self.data_dir + "/info/nonltr.gff3")
cmd1 = self.base_path + "/nonltr/toGFF.py %(nonltr_out_path)s %(nonltr_gff_path)s"
res1 = self.run_cmd(cmd1)
end = time.time()
print ('nonltr: finishing (elapsed time: {0} secs)'.format(int(round(end -
start))))
def forward_strand(self):
mypath = self.plus_dir
out_dir = self.plus_out_dir
for (dirpath, dirnames, filenames) in os.walk(mypath):
break
for name in filenames:
file_path = utils.get_abspath(dirpath + "/" + name)
# Rename to sequence id
sid = getid(file_path)
new_path = utils.get_abspath(dirpath + "/" + sid)
os.rename(file_path, new_path)
command = self.cmd_hmm + (" --dna=%s --out=%s --hmmerv=%s" %
(new_path, out_dir, self.hmmerv))
command = command.split()
self.processes.add(Popen(command, stdout=PIPE,
stderr=PIPE))
if len(self.processes) >= self.max_processes:
time.sleep(.1)
self.processes.difference_update([p for p in self.processes if
p.poll() is not None])
#print dirpath, dirnames, filenames
for p in self.processes:
if p.poll() is None:
p.wait()
self.post_processing_after_forward_strand()
def nonltr(self):
print 'nonltr: starting'
start = time.time()
# nonltr
#cmd0 = self.base_path + "/nonltr/run_MGEScan.pl \
# -genome=%(genome_dir)s \
# -data=%(data_dir)s \
# -hmmerv=%(hmmerv)s"
cmd0 = "python " + self.base_path + "/nonltr/nonltr.py " + \
"%(genome_dir)s " + \
"%(data_dir)s "
if self.mpi_enabled:
#cmd0 = (cmd0 + " -mpi=%(mpi_enabled)s")
cmd0 = (cmd0 + " --mpi=%(mpi_enabled)s")
res0 = self.run_cmd(cmd0)
# gff3
self.nonltr_out_path = utils.get_abspath(self.data_dir + "/info/full/")
self.nonltr_gff_path = utils.get_abspath(self.data_dir +
"/info/nonltr.gff3")
cmd1 = self.base_path + "/nonltr/toGFF.py %(nonltr_out_path)s %(nonltr_gff_path)s"
res1 = self.run_cmd(cmd1)
end = time.time()
print('nonltr: finishing (elapsed time: {0} secs)'.format(
int(round(end - start))))
def backward_strand(self):
mypath = self.minus_dir
out_dir = self.minus_out_dir
for (dirpath, dirnames, filenames) in os.walk(mypath):
break
for name in filenames:
file_path = utils.get_abspath(dirpath + "/" + name)
# Rename to sequence id
sid = getid(file_path)
new_path = utils.get_abspath(dirpath + "/" + sid)
os.rename(file_path, new_path)
command = self.cmd_hmm + (" --dna=%s --out=%s --hmmerv=%s" %
(new_path, out_dir, self.hmmerv))
command = command.split()
self.processes.add(Popen(command, stdout=PIPE, stderr=PIPE))
if len(self.processes) >= self.max_processes:
time.sleep(.1)
self.processes.difference_update(
[p for p in self.processes if p.poll() is not None])
#print dirpath, dirnames, filenames
for p in self.processes:
if p.poll() is None:
p.wait()
self.post_processing_after_reverse_strand()
def post_processing(self, out_dir, dir, reverse_yn):
utils.silentremove(utils.get_abspath(out_dir + "/out1/aaaaa"))
utils.silentremove(utils.get_abspath(out_dir + "out1/bbbbb"))
utils.silentremove(utils.get_abspath(out_dir + "out1/ppppp"))
utils.silentremove(utils.get_abspath(out_dir + "out1/qqqqq"))
cmd = self.cmd_post_process + (" --dna=%s --out=%s --rev=%s" %
(dir, out_dir, reverse_yn))
self.run_cmd(cmd)
def post_processing(self, out_dir, dir, reverse_yn):
utils.silentremove(utils.get_abspath(out_dir + "/out1/aaaaa"))
utils.silentremove(utils.get_abspath(out_dir + "out1/bbbbb"))
utils.silentremove(utils.get_abspath(out_dir + "out1/ppppp"))
utils.silentremove(utils.get_abspath(out_dir + "out1/qqqqq"))
cmd = self.cmd_post_process + (" --dna=%s --out=%s --rev=%s" %
(dir, out_dir, reverse_yn))
self.run_cmd(cmd)
def set_inputs(self):
self.data_dir = utils.get_abspath(self.args['--output'])
self.genome_dir = utils.get_abspath(self.args['<genome_dir>'])
self.ltr_enabled = self.args['ltr']
self.nonltr_enabled = self.args['nonltr']
self.mpi_enabled = self.args['--mpi']
if(self.mpi_enabled and not self.ltr_enabled and not self.nonltr_enabled):
self.mpi_enabled = str(int(math.ceil(1.0*int(self.args['--mpi'])/2)))
self.debug = self.args['--debug']
def toGFF(self):
if self.gff3_enabled:
# Assume info is a only directory in genome_dir
shutil.move(self.genome_dir + "/info", self.data_dir)
# gff3
self.nonltr_out_path = utils.get_abspath(self.data_dir + "/info/full/")
self.nonltr_gff_path = utils.get_abspath(self.data_dir + "/info/nonltr.gff3")
cmd = self.cmd_togff + " %(nonltr_out_path)s %(nonltr_gff_path)s"
res = self.run_cmd(cmd)
def set_inputs(self):
self.data_dir = utils.get_abspath(self.args['--output'])
self.genome_dir = utils.get_abspath(self.args['<genome_dir>'])
self.ltr_enabled = self.args['ltr']
self.nonltr_enabled = self.args['nonltr']
self.mpi_enabled = self.args['--mpi']
if (self.mpi_enabled and not self.ltr_enabled
and not self.nonltr_enabled):
self.mpi_enabled = str(
int(math.ceil(1.0 * int(self.args['--mpi']) / 2)))
self.debug = self.args['--debug']
def toGFF(self):
if self.gff3_enabled:
# Assume info is a only directory in genome_dir
shutil.move(self.genome_dir + "/info", self.data_dir)
# gff3
self.nonltr_out_path = utils.get_abspath(self.data_dir + "/info/full/")
self.nonltr_gff_path = utils.get_abspath(self.data_dir +
"/info/nonltr.gff3")
cmd = self.cmd_togff + " %(nonltr_out_path)s %(nonltr_gff_path)s"
res = self.run_cmd(cmd)
def set_defaults(self):
"""Set default values to run programs
For LTR,
min_dist: minimum distance(bp) between LTRs.
max_dist: maximum distance(bp) between LTRS
min_len_ltr: minimum length(bp) of LTR.
max_len_ltr: maximum length(bp) of LTR.
ltr_sim_condition: minimum similarity(%) for LTRs in an element.
cluster_sim_condition: minimum similarity(%) for LTRs in a cluster
len_condition: minimum length(bp) for LTRs aligned in local alignment.
"""
if self.data_dir:
self.data_dir = utils.create_directory(self.data_dir, False)
else:
self.data_dir = \
utils.create_directory(utils.get_abspath(self.default_output_path))
self.hmmerv = 3
self.min_dist = 2000
self.max_dist = 20000
self.min_len_ltr = 130
self.max_len_ltr = 2000
self.ltr_sim_condition = 70
self.cluster_sim_condition = 70
self.len_condition = 70
self.sw_rm = "No" # or Yes
self.scaffold = "" # or directory
def set_defaults(self):
"""Set default values to run programs
For LTR,
min_dist: minimum distance(bp) between LTRs.
max_dist: maximum distance(bp) between LTRS
min_len_ltr: minimum length(bp) of LTR.
max_len_ltr: maximum length(bp) of LTR.
ltr_sim_condition: minimum similarity(%) for LTRs in an element.
cluster_sim_condition: minimum similarity(%) for LTRs in a cluster
len_condition: minimum length(bp) for LTRs aligned in local alignment.
"""
if self.data_dir:
self.data_dir = utils.create_directory(self.data_dir, False)
else:
self.data_dir = \
utils.create_directory(utils.get_abspath(self.default_output_path))
self.hmmerv = 3
self.min_dist = 2000
self.max_dist = 20000
self.min_len_ltr = 130
self.max_len_ltr = 2000
self.ltr_sim_condition = 70
self.cluster_sim_condition = 70
self.len_condition = 70
self.sw_rm = "No" # or Yes
self.scaffold = "" # or directory
def reverse_complement(self, directory):
mypath = self.genome_path
for (dirpath, dirnames, filenames) in os.walk(mypath):
break
utils.create_directory(directory, False)
for name in filenames:
file_path = utils.get_abspath(dirpath + "/" + name)
reverse_complement_fasta(file_path, directory)
def reverse_complement(self, directory):
mypath = self.genome_path
for (dirpath, dirnames, filenames) in os.walk(mypath):
break
utils.create_directory(directory, False)
for name in filenames:
file_path = utils.get_abspath(dirpath + "/" + name)
reverse_complement_fasta(file_path, directory)
def reverse_complement(self):
mypath = self.genome_dir
for (dirpath, dirnames, filenames) in os.walk(mypath):
break
directory = self.minus_dir
if not os.path.exists(directory):
os.makedirs(directory)
for name in filenames:
file_path = utils.get_abspath(dirpath + "/" + name)
reverse_complement_fasta(file_path, directory)
def reverse_complement(self):
mypath = self.genome_dir
for (dirpath, dirnames, filenames) in os.walk(mypath):
break
directory = self.minus_dir
if not os.path.exists(directory):
os.makedirs(directory)
for name in filenames:
file_path = utils.get_abspath(dirpath + "/" + name)
reverse_complement_fasta(file_path, directory)
def ltr(self):
print 'ltr: starting'
start = time.time()
# scaffold
# repeatmasker
cmd0 = self.base_path + "/ltr/pre_process.pl \
-genome=%(genome_dir)s \
-data=%(data_dir)s \
-sw_rm=%(sw_rm)s \
-scaffold=%(scaffold)s"
res0 = self.run_cmd(cmd0)
# find-ltr
cmd1 = self.base_path + "/ltr/find_ltr.pl \
-genome=%(genome_dir)s \
-data=%(data_dir)s \
-hmmerv=%(hmmerv)s \
-min_dist=%(min_dist)s \
-max_dist=%(max_dist)s \
-min_len_ltr=%(min_len_ltr)s \
-max_len_ltr=%(max_len_ltr)s \
-ltr_sim_condition=%(ltr_sim_condition)s \
-cluster_sim_condition=%(cluster_sim_condition)s \
-len_condition=%(len_condition)s"
if self.mpi_enabled:
cmd1 = (cmd1 + " -mpi=%(mpi_enabled)s")
res1 = self.run_cmd(cmd1)
# gff3
self.ltr_out_path = utils.get_abspath(self.data_dir + "/ltr/ltr.out")
self.ltr_gff_path = utils.get_abspath(self.data_dir + "/ltr/ltr.gff3")
cmd2 = self.base_path + "/ltr/toGFF.py %(ltr_out_path)s %(ltr_gff_path)s"
res2 = self.run_cmd(cmd2)
end = time.time()
print('ltr: finishing (elapsed time: {0} secs)'.format(
int(round(end - start))))
def ltr(self):
print 'ltr: starting'
start = time.time()
# scaffold
# repeatmasker
cmd0 = self.base_path + "/ltr/pre_process.pl \
-genome=%(genome_dir)s \
-data=%(data_dir)s \
-sw_rm=%(sw_rm)s \
-scaffold=%(scaffold)s"
res0 = self.run_cmd(cmd0)
# find-ltr
cmd1 = self.base_path + "/ltr/find_ltr.pl \
-genome=%(genome_dir)s \
-data=%(data_dir)s \
-hmmerv=%(hmmerv)s \
-min_dist=%(min_dist)s \
-max_dist=%(max_dist)s \
-min_len_ltr=%(min_len_ltr)s \
-max_len_ltr=%(max_len_ltr)s \
-ltr_sim_condition=%(ltr_sim_condition)s \
-cluster_sim_condition=%(cluster_sim_condition)s \
-len_condition=%(len_condition)s"
if self.mpi_enabled:
cmd1 = (cmd1 + " -mpi=%(mpi_enabled)s")
res1 = self.run_cmd(cmd1)
# gff3
self.ltr_out_path = utils.get_abspath(self.data_dir + "/ltr/ltr.out")
self.ltr_gff_path = utils.get_abspath(self.data_dir + "/ltr/ltr.gff3")
cmd2 = self.base_path + "/ltr/toGFF.py %(ltr_out_path)s %(ltr_gff_path)s"
res2 = self.run_cmd(cmd2)
end = time.time()
print ('ltr: finishing (elapsed time: {0} secs)'.format(int(round(end -
start))))
def set_input(self, path):
self.input_file = utils.get_abspath(path)
return self.input_file
def set_inputs(self):
self.data_dir = utils.get_abspath(self.args['--output'])
self.genome_dir = utils.get_abspath(self.args['<genome_dir>'])
def set_inputs(self):
self.data_dir = utils.get_abspath(self.args['--output'])
self.genome_dir = utils.get_abspath(self.args['<genome_dir>'])
def set_output(self, path):
self.result_path = utils.get_abspath(path)
return self.result_path
def set_inputs(self):
self.args = docopt(self.__doc__, version=self.ver)
self.nmpi = self.args['--mpi']
self.genome_path = utils.get_abspath(self.args['<input>'])
self.output_path = utils.get_abspath(self.args['<output>'])
def set_inputs(self):
self.args = docopt(self.__doc__, version=self.ver)
self.nmpi = self.args['--mpi']
self.genome_path = utils.get_abspath(self.args['<input>'])
self.output_path = utils.get_abspath(self.args['<output>'])
def set_datadir(self, path): self.datadir = utils.get_abspath(path) return self.datadir