def merge_subalignments(self):
'''
Merge alignment subset extended alignments to get one extended alignment
for current placement subset.
'''
pp = self.placement_problem
_LOG.info("Merging sub-alignments for placement problem : %s." %(pp.label))
''' First assign fragments to the placement problem'''
pp.fragments = pp.parent.fragments.get_soft_sub_alignment([])
for ap in pp.get_children():
pp.fragments.seq_names |= set(ap.fragments)
''' Then Build an extended alignment by merging all hmmalign results'''
extendedAlignment = ExtendedAlignment(pp.fragments.seq_names)
for ap in pp.children:
assert isinstance(ap, SeppProblem)
''' Get all fragment chunk alignments for this alignment subset'''
aligned_files = [fp.get_job_result_by_name('hmmalign') for
fp in ap.children if
fp.get_job_result_by_name('hmmalign') is not None]
_LOG.info("Merging fragment chunks for subalignment : %s." %(ap.label))
ap_alg = ap.read_extendend_alignment_and_relabel_columns\
(ap.jobs["hmmbuild"].infile , aligned_files)
_LOG.info("Merging alignment subset into placement subset: %s." %(ap.label))
extendedAlignment.merge_in(ap_alg,convert_to_string=False)
del ap_alg
extendedAlignment.from_bytearray_to_string()
return extendedAlignment
def merge_subalignments(self):
'''
Merge alignment subset extended alignments to get one extended alignment
for current placement subset.
'''
pp = self.placement_problem
_LOG.info("Merging sub-alignments for placement problem : %s." %
(pp.label))
''' First assign fragments to the placement problem'''
pp.fragments = pp.parent.fragments.get_soft_sub_alignment([])
for ap in pp.get_children():
pp.fragments.seq_names |= set(ap.fragments)
''' Then Build an extended alignment by merging all hmmalign results'''
extendedAlignment = ExtendedAlignment(pp.fragments.seq_names)
for ap in pp.children:
assert isinstance(ap, SeppProblem)
''' Get all fragment chunk alignments for this alignment subset'''
aligned_files = [
fp.get_job_result_by_name('hmmalign') for fp in ap.children
if fp.get_job_result_by_name('hmmalign') is not None
]
_LOG.info("Merging fragment chunks for subalignment : %s." %
(ap.label))
ap_alg = ap.read_extendend_alignment_and_relabel_columns\
(ap.jobs["hmmbuild"].infile , aligned_files)
_LOG.info("Merging alignment subset into placement subset: %s." %
(ap.label))
extendedAlignment.merge_in(ap_alg, convert_to_string=False)
del ap_alg
extendedAlignment.from_bytearray_to_string()
return extendedAlignment
def merge_subalignments(self):
'''
Merge alignment subset extended alignments to get one extended
alignment for current placement subset.
'''
pp = self.placement_problem
_LOG.info("Merging sub-alignments for placement problem : %s." %
(pp.label))
''' First find fragments assigned to this placement problem'''
pp.fragments = pp.parent.fragments.get_soft_sub_alignment([])
for ap in pp.get_children():
pp.fragments.seq_names |= set(ap.fragments)
''' Then, gather a list of all alignments relevant to this placement
subset'''
fragfilesperap = dict()
for ap in pp.children:
assert isinstance(ap, SeppProblem)
''' Get all fragment chunk alignments for this alignment subset'''
aligned_files = [fp.get_job_result_by_name('hmmalign') for
fp in ap.children]
fragfilesperap[ap] = aligned_files
''' Now, build an extended alignment *per each fragment chunk*.
Simply merge all hmmalign results for fragment chunk numbered i'''
extendedAlignments = []
for i in range(0, self.root_problem.fragment_chunks):
extendedAlignment = ExtendedAlignment(pp.fragments.seq_names)
for ap in pp.children:
# _LOG.debug("Merging fragment chunks for subalignment : %s."
# %(ap.label))
if fragfilesperap[ap][i]:
ap_alg = ap.read_extendend_alignment_and_relabel_columns(
ap.jobs["hmmbuild"].infile, [fragfilesperap[ap][i]])
else:
ap_alg = ap.read_extendend_alignment_and_relabel_columns(
ap.jobs["hmmbuild"].infile, [])
_LOG.debug(
("Merging alignment subset into placement subset for "
"chunk %d: %s.") % (i, ap.label))
extendedAlignment.merge_in(ap_alg, convert_to_string=False)
'''Extended alignmnts have all fragments. remove the ones that
don't belong to thsi chunk'''
extendedAlignment.remove_missing_fragments()
extendedAlignment.from_bytearray_to_string()
extendedAlignments.append(extendedAlignment)
return extendedAlignments
def merge_subalignments(self):
'''
Merge alignment subset extended alignments to get one extended
alignment for current placement subset.
'''
pp = self.placement_problem
_LOG.info("Merging sub-alignments for placement problem : %s." %
(pp.label))
''' First find fragments assigned to this placement problem'''
pp.fragments = pp.parent.fragments.get_soft_sub_alignment([])
for ap in pp.get_children():
pp.fragments.seq_names |= set(ap.fragments)
''' Then, gather a list of all alignments relevant to this placement
subset'''
fragfilesperap = dict()
for ap in pp.children:
assert isinstance(ap, SeppProblem)
''' Get all fragment chunk alignments for this alignment subset'''
aligned_files = [
fp.get_job_result_by_name('hmmalign') for fp in ap.children
]
fragfilesperap[ap] = aligned_files
''' Now, build an extended alignment *per each fragment chunk*.
Simply merge all hmmalign results for fragment chunk numbered i'''
extendedAlignments = []
for i in range(0, self.root_problem.fragment_chunks):
extendedAlignment = ExtendedAlignment(pp.fragments.seq_names)
for ap in pp.children:
# _LOG.debug("Merging fragment chunks for subalignment : %s."
# %(ap.label))
if fragfilesperap[ap][i]:
ap_alg = ap.read_extendend_alignment_and_relabel_columns(
ap.jobs["hmmbuild"].infile, [fragfilesperap[ap][i]])
else:
ap_alg = ap.read_extendend_alignment_and_relabel_columns(
ap.jobs["hmmbuild"].infile, [])
_LOG.debug(
("Merging alignment subset into placement subset for "
"chunk %d: %s.") % (i, ap.label))
extendedAlignment.merge_in(ap_alg, convert_to_string=False)
'''Extended alignmnts have all fragments. remove the ones that
don't belong to thsi chunk'''
extendedAlignment.remove_missing_fragments()
extendedAlignment.from_bytearray_to_string()
extendedAlignments.append(extendedAlignment)
return extendedAlignments
def merge_results(self):
assert \
len(self.root_problem.get_children()) == 1, \
"Currently UPP works with only one placement subset."
'''
Merge alignment subset extended alignments to get one extended
alignment for current placement subset.
'''
pp = self.root_problem.get_children()[0]
_LOG.info("Merging sub-alignments for placement problem : %s." %
pp.label)
''' First assign fragments to the placement problem'''
pp.fragments = pp.parent.fragments.get_soft_sub_alignment([])
for ap in pp.get_children():
pp.fragments.seq_names |= set(ap.fragments)
''' Then Build an extended alignment by merging all hmmalign results'''
_LOG.debug("fragments are %d:\n %s" %
(len(pp.fragments.seq_names), pp.fragments.seq_names))
extendedAlignment = ExtendedAlignment(pp.fragments.seq_names)
for ap in pp.children:
assert isinstance(ap, SeppProblem)
''' Get all fragment chunk alignments for this alignment subset'''
aligned_files = [
fp.get_job_result_by_name('hmmalign') for fp in ap.children
if fp.get_job_result_by_name('hmmalign') is not None
]
_LOG.debug("Merging fragment chunks for subalignment : %s." %
ap.label)
ap_alg = ap.read_extendend_alignment_and_relabel_columns(
ap.jobs["hmmbuild"].infile, aligned_files)
_LOG.debug("Merging alignment subset into placement subset: %s." %
ap.label)
extendedAlignment.merge_in(ap_alg, convert_to_string=False)
extendedAlignment.from_bytearray_to_string()
self.results = extendedAlignment