def create_job(self, alignment, guide_tree=None, **kwargs):
job_id = kwargs.get('context_str', '') + '_prank'
if alignment.get_num_taxa() == 0:
return FakeJob(alignment, context_str=job_id)
new_alignment = alignment.unaligned()
if new_alignment.get_num_taxa() < 2:
return FakeJob(new_alignment, context_str=job_id)
scratch_dir, seqfn, alignedfn = self._prepare_input(
new_alignment, **kwargs)
invoc = [
self.exe, '-once', '-noxml', '-notree', '-nopost', '+F', '-quiet',
'-matinitsize=5', '-uselogs',
'-d=%s' % seqfn,
'-o=%s' % alignedfn
]
invoc.extend(self.user_opts)
alignedfn = alignedfn + '.1.fas'
return self._finish_standard_job(alignedfn=alignedfn,
datatype=alignment.datatype,
invoc=invoc,
scratch_dir=scratch_dir,
job_id=job_id,
delete_temps=kwargs.get(
'delete_temps',
self.delete_temps))
def create_job(self, alignment, guide_tree=None, **kwargs):
job_id = kwargs.get('context_str', '') + '_clustalw2'
if alignment.get_num_taxa() == 0:
return FakeJob(alignment, context_str=job_id)
new_alignment = alignment.unaligned()
if new_alignment.get_num_taxa() < 2:
return FakeJob(new_alignment, context_str=job_id)
scratch_dir, seqfn, alignedfn = self._prepare_input(
new_alignment, **kwargs)
invoc = [
self.exe, '-align',
'-infile=%s' % seqfn,
'-outfile=%s' % alignedfn, '-output=fasta'
]
invoc.extend(self.user_opts)
return self._finish_standard_job(alignedfn=alignedfn,
datatype=alignment.datatype,
invoc=invoc,
scratch_dir=scratch_dir,
job_id=job_id,
delete_temps=kwargs.get(
'delete_temps',
self.delete_temps))
def create_job(self, alignment, guide_tree=None, **kwargs):
job_id = kwargs.get('context_str', '') + '_probcons'
if alignment.get_num_taxa() == 0:
return FakeJob(alignment, context_str=job_id)
new_alignment = alignment.unaligned()
if new_alignment.get_num_taxa() < 2:
return FakeJob(new_alignment, context_str=job_id)
scratch_dir, seqfn, alignedfn = self._prepare_input(
new_alignment, **kwargs)
invoc = [self.exe, seqfn]
invoc.extend(self.user_opts)
# The probcons job creation is slightly different from the other
# aligners because we redirect and read standard output.
return self._finish_standard_job(alignedfn=alignedfn,
datatype=alignment.datatype,
invoc=invoc,
scratch_dir=scratch_dir,
job_id=job_id,
delete_temps=kwargs.get(
'delete_temps',
self.delete_temps),
stdout=alignedfn)
def create_job(self, alignment, guide_tree=None, **kwargs):
job_id = kwargs.get('context_str', '') + '_ginsi'
if alignment.get_num_taxa() == 0:
return FakeJob(alignment, context_str=job_id)
new_alignment = alignment.unaligned()
if new_alignment.get_num_taxa() < 2:
return FakeJob(new_alignment, context_str=job_id)
scratch_dir, seqfn, alignedfn = self._prepare_input(
new_alignment, **kwargs)
invoc = []
if platform.system() == "Windows":
invoc.append(self.exe)
else:
invoc.extend([self.exe])
invoc.extend(['--globalpair', '--maxiterate', '1000'])
if '--ep' not in self.user_opts:
invoc.extend(['--ep', '0.123'])
invoc.extend(['--quiet'])
invoc.extend(self.user_opts)
invoc.extend(['--thread', str(kwargs.get('num_cpus', 1))])
invoc.append(seqfn)
# The MAFFT job creation is slightly different from the other
# aligners because we redirect and read standard output.
return self._finish_standard_job(alignedfn=alignedfn,
datatype=alignment.datatype,
invoc=invoc,
scratch_dir=scratch_dir,
job_id=job_id,
delete_temps=kwargs.get(
'delete_temps',
self.delete_temps),
stdout=alignedfn)
def create_job(self, alignment, guide_tree=None, **kwargs):
job_id = kwargs.get('context_str', '') + '_opal'
if alignment.get_num_taxa() == 0:
return FakeJob(alignment, context_str=job_id)
new_alignment = alignment.unaligned()
if new_alignment.get_num_taxa() < 2:
return FakeJob(new_alignment, context_str=job_id)
scratch_dir, seqfn, alignedfn = self._prepare_input(
new_alignment, **kwargs)
invoc = [
'java',
'-Xmx%dm' % self.max_mem_mb, '-jar', self.exe, '--in', seqfn,
'--out', alignedfn, '--quiet'
]
invoc.extend(self.user_opts)
return self._finish_standard_job(alignedfn=alignedfn,
datatype=alignment.datatype,
invoc=invoc,
scratch_dir=scratch_dir,
job_id=job_id,
delete_temps=kwargs.get(
'delete_temps',
self.delete_temps))
def create_job(self, alignment, guide_tree=None, **kwargs):
job_id = kwargs.get('context_str', '') + '_baliphyalign'
#TODO: figure out why time limit is not being passed in here...
# time_to_run = kwargs.get('time_limit','12')
time_to_run = '0.25' #DEBUG ONLY
if alignment.get_num_taxa() == 0:
return FakeJob(alignment, context_str=job_id)
new_alignment = alignment.unaligned()
if new_alignment.get_num_taxa() < 2:
return FakeJob(new_alignment, context_str=job_id)
scratch_dir, seqfn, alignedfn = self._prepare_input(
new_alignment, **kwargs)
invoc = [self.exe, seqfn, scratch_dir, time_to_run, alignedfn]
invoc.extend(self.user_opts)
# print ' '.join(invoc)
return self._finish_standard_job(alignedfn=alignedfn,
datatype=alignment.datatype,
invoc=invoc,
scratch_dir=scratch_dir,
job_id=job_id,
delete_temps=kwargs.get(
'delete_temps',
self.delete_temps))
def create_job(self, alignment, guide_tree=None, **kwargs):
job_id = kwargs.get('context_str', '') + '_mafft'
if alignment.get_num_taxa() == 0:
return FakeJob(alignment, context_str=job_id)
new_alignment = alignment.unaligned()
if new_alignment.get_num_taxa() < 2:
return FakeJob(new_alignment, context_str=job_id)
scratch_dir, seqfn, alignedfn = self._prepare_input(
new_alignment, **kwargs)
invoc = []
if platform.system() == "Windows":
invoc.append(self.exe)
else:
invoc.extend([self.exe])
if len(alignment) <= 200 and new_alignment.max_sequence_length(
) < 50000:
invoc.extend(['--localpair', '--maxiterate', '1000'])
if '--ep' not in self.user_opts:
invoc.extend(['--ep', '0.123'])
invoc.extend(['--quiet'])
invoc.extend(self.user_opts)
if get_sparkcontext():
#import multiprocessing
#available_cpus = multiprocessing.cpu_count()
MESSENGER.send_info("Using spark to launch MAFFT")
invoc.extend(['--thread', str(kwargs.get('num_cpus_spark', 1))])
#invoc.extend(['--thread', str(kwargs.get('num_cpus_spark', available_cpus))])
else:
import multiprocessing
available_cpus = multiprocessing.cpu_count()
MESSENGER.send_info("NOT using spark to launch MAFFT")
invoc.extend(
['--thread',
str(kwargs.get('num_cpus', available_cpus))])
invoc.append(seqfn)
# The MAFFT job creation is slightly different from the other
# aligners because we redirect and read standard output.
return self._finish_standard_job(alignedfn=alignedfn,
datatype=alignment.datatype,
invoc=invoc,
scratch_dir=scratch_dir,
job_id=job_id,
delete_temps=kwargs.get(
'delete_temps',
self.delete_temps),
stdout=alignedfn)
def create_job(self, alignment1, alignment2, **kwargs):
job_id = kwargs.get('context_str', '') + '_opal'
if (alignment1.get_num_taxa() < 1) or (alignment2.get_num_taxa() < 1):
alignment1.update(alignment2)
return FakeJob(alignment1, context_str=job_id)
scratch_dir, seqfn1, seqfn2, outfn = self._prepare_input(
alignment1, alignment2, **kwargs)
assert (alignment1.datatype == alignment2.datatype)
invoc = [
'java',
'-Xmx%dm' % self.max_mem_mb, '-jar', self.exe, '--in', seqfn1,
'--in2', seqfn2, '--out', outfn, '--align_method', 'profile'
]
# invoc = ['java', '-Xmx10000m', '-jar', self.exe, '--in', seqfn1, '--in2', seqfn2, '--out', outfn, '--align_method', 'profile']
invoc.extend(self.user_opts)
return self._finish_standard_job(alignedfn=outfn,
datatype=alignment1.datatype,
invoc=invoc,
scratch_dir=scratch_dir,
job_id=job_id,
delete_temps=kwargs.get(
'delete_temps',
self.delete_temps))
def create_job(self, alignment, guide_tree=None, **kwargs):
job_id = kwargs.get('context_str', '') + '_padaligner'
if alignment.get_num_taxa() == 0:
return FakeJob(alignment, context_str=job_id)
new_alignment = alignment.unaligned()
if new_alignment.get_num_taxa() < 2:
return FakeJob(new_alignment, context_str=job_id)
scratch_dir, seqfn, alignedfn = self._prepare_input(
new_alignment, **kwargs)
invoc = [
sys.executable, self.exe, alignment.datatype, seqfn, alignedfn
]
return self._finish_standard_job(alignedfn=alignedfn,
datatype=alignment.datatype,
invoc=invoc,
scratch_dir=scratch_dir,
job_id=job_id,
delete_temps=kwargs.get(
'delete_temps',
self.delete_temps))
def create_job(self,
alignment,
starting_tree=None,
name='default',
**kwargs):
assert (starting_tree)
job_id = kwargs.get('context_str', '') + '_fake'
if isinstance(starting_tree, str):
tree_str = starting_tree
else:
tree_str = str(starting_tree)
score = hash(tree_str) / 10000.0
blob = (score, tree_str)
return FakeJob(blob, context_str=job_id)
def create_job(self, alignment=None, guide_tree=None, **kwargs):
job_id = kwargs.get('context_str', '') + '_customaligner'
if alignment.get_num_taxa() == 0:
return FakeJob(alignment, context_str=job_id)
new_alignment = alignment.unaligned()
if new_alignment.get_num_taxa() < 2:
return FakeJob(new_alignment, context_str=job_id)
scratch_dir, seqfn, alignedfn = self._prepare_input(
new_alignment, **kwargs)
# rpc = lambda : read_internal_alignment(alignedfn,
# datatype=alignment.datatype)
# self.scratch_dir=scratch_dir
# self.seqfn=seqfn
# self.alignedfn=alignedfn
return FakeJob(file_read_job=True,
context_str=job_id,
scratch_dir=scratch_dir,
seqfn=seqfn,
alignedfn=alignedfn,
datatype=alignment.datatype)
def create_job(self, alignment1, alignment2, **kwargs):
job_id = kwargs.get('context_str', '') + '_padmerger'
if (alignment1.get_num_taxa() < 1) or (alignment2.get_num_taxa() < 1):
alignment1.update(alignment2)
return FakeJob(alignment1, context_str=job_id)
scratch_dir, seqfn1, seqfn2, outfn = self._prepare_input(
alignment1, alignment2, **kwargs)
invoc = [
sys.executable, self.exe, alignment1.datatype, seqfn1, seqfn2,
outfn
]
return self._finish_standard_job(alignedfn=outfn,
datatype=alignment1.datatype,
invoc=invoc,
scratch_dir=scratch_dir,
job_id=job_id,
delete_temps=kwargs.get(
'delete_temps',
self.delete_temps))
def create_job(self, alignment1, alignment2, **kwargs):
job_id = kwargs.get('context_str', '') + '_muscle'
if (alignment1.get_num_taxa() < 1) or (alignment2.get_num_taxa() < 1):
alignment1.update(alignment2)
return FakeJob(alignment1, context_str=job_id)
scratch_dir, seqfn1, seqfn2, outfn = self._prepare_input(
alignment1, alignment2, **kwargs)
invoc = [
self.exe, '-in1', seqfn1, '-in2', seqfn2, '-out', outfn, '-quiet',
'-profile'
]
invoc.extend(self.user_opts)
return self._finish_standard_job(alignedfn=outfn,
datatype=alignment1.datatype,
invoc=invoc,
scratch_dir=scratch_dir,
job_id=job_id,
delete_temps=kwargs.get(
'delete_temps',
self.delete_temps))
def create_job(self, alignment1, alignment2, **kwargs):
alignment1.update(alignment2)
job_id = kwargs.get('context_str', '') + '_fakemerger'
return FakeJob(alignment1, context_str=job_id)
def create_job(self, alignment, guide_tree=None, **kwargs):
job_id = kwargs.get('context_str', '') + '_fakealigner'
return FakeJob(alignment, context_str=job_id)
def create_file_read_job(self, **kwargs):
return FakeJob(**kwargs)