def test1000T(self):
sd = SequenceDataset()
fp = data_source_path('1000T.fasta')
sd.read(open(fp, 'rU'), file_format='FASTA', datatype='DNA')
fp = data_source_path('1000T.tree')
tree_list = read_and_encode_splits(sd.dataset, open(fp, "rU"))
self.assertEqual(len(tree_list), 1)
def test1000T(self):
sd = SequenceDataset()
fp = data_source_path('1000T.fasta')
sd.read(open(fp, 'rU'), file_format='FASTA', datatype='DNA')
fp = data_source_path('1000T.tree')
tree_list = read_and_encode_splits(sd.dataset, open(fp, "rU"))
self.assertEqual(len(tree_list), 1)
def testCentroidEdge(self):
sd = SequenceDataset()
fp = data_source_path('100T.fasta')
sd.read(open(fp, 'rU'), file_format='FASTA', datatype='DNA')
fp = data_source_path('100T.tree')
tree_list = read_and_encode_splits(sd.dataset, open(fp, "rU"))
self.assertEqual(len(tree_list), 1)
t = PhylogeneticTree(tree_list[0])
self._do_test_centroid(t)
def testCentroidEdge(self):
sd = SequenceDataset()
fp = data_source_path('100T.fasta')
sd.read(open(fp, 'rU'), file_format='FASTA', datatype='DNA')
fp = data_source_path('100T.tree')
tree_list = read_and_encode_splits(sd.dataset, open(fp, "rU"))
self.assertEqual(len(tree_list), 1)
t = PhylogeneticTree(tree_list[0])
self._do_test_centroid(t)
def pasta_prelim_step1(multilocus_dataset, temporaries_dir, user_config):
'''
Broadly speaking, this subroutine starts the threads, reads the tree andinitializes some variables.
'''
options = user_config.commandline
user_config.save_to_filepath(os.path.join(temporaries_dir, 'last_used.cfg'))
if options.timesfile:
f = open_with_intermediates(options.timesfile, 'a')
f.close()
set_timing_log_filepath(options.timesfile)
############################################################################
# Launch threads to do work
#####
pasta_config = user_config.get("sate")
start_worker(pasta_config.num_cpus)
# _LOG.debug("start reading the input alignment")
# multilocus_dataset = read_input_sequences(user_config.input_seq_filepaths,
# datatype=user_config.commandline.datatype,
# missing=user_config.commandline.missing)
############################################################################
# We must read the incoming tree in before we call the get_sequences_for_pasta
# function that relabels that taxa in the dataset
######
alignment_as_tmp_filename_to_report = None
tree_as_tmp_filename_to_report = None
starting_tree = None
tree_file = options.treefile
if tree_file:
if not os.path.exists(tree_file):
raise Exception('The tree file "%s" does not exist' % tree_file)
tree_f = open(tree_file, 'rU')
MESSENGER.send_info('Reading starting trees from "%s"...' % tree_file)
try:
tree_list = read_and_encode_splits(multilocus_dataset.dataset, tree_f,
starting_tree=True)
except KeyError:
MESSENGER.send_error(
"Error in reading the treefile, probably due to a name in the tree that does not match the names in the input sequence files.\n")
raise
except:
MESSENGER.send_error("Error in reading the treefile.\n")
raise
tree_f.close()
if len(tree_list) > 1:
MESSENGER.send_warning(
'%d starting trees found in "%s". The first tree will be used.' % (len(tree_list), tree_file))
starting_tree = tree_list[0]
score = None
tree_as_tmp_filename_to_report = tree_file
else:
score=None
return alignment_as_tmp_filename_to_report, options, pasta_config, score, starting_tree, tree_as_tmp_filename_to_report, tree_file
def finish_pasta_execution(pasta_team, user_config, temporaries_dir,
pasta_products, multilocus_dataset):
global _RunningJobs
options = user_config.commandline
user_config.save_to_filepath(os.path.join(temporaries_dir,
'last_used.cfg'))
if options.timesfile:
f = open_with_intermediates(options.timesfile, 'a')
f.close()
set_timing_log_filepath(options.timesfile)
############################################################################
# Launch threads to do work
#####
pasta_config = user_config.get("sate")
start_worker(pasta_config.num_cpus)
#_LOG.debug("start reading the input alignment")
#multilocus_dataset = read_input_sequences(user_config.input_seq_filepaths,
# datatype=user_config.commandline.datatype,
# missing=user_config.commandline.missing)
############################################################################
# We must read the incoming tree in before we call the get_sequences_for_pasta
# function that relabels that taxa in the dataset
######
alignment_as_tmp_filename_to_report = None
tree_as_tmp_filename_to_report = None
starting_tree = None
tree_file = options.treefile
if tree_file:
if not os.path.exists(tree_file):
raise Exception('The tree file "%s" does not exist' % tree_file)
tree_f = open(tree_file, 'rU')
MESSENGER.send_info('Reading starting trees from "%s"...' % tree_file)
try:
tree_list = read_and_encode_splits(multilocus_dataset.dataset,
tree_f,
starting_tree=True)
except KeyError:
MESSENGER.send_error(
"Error in reading the treefile, probably due to a name in the tree that does not match the names in the input sequence files.\n"
)
raise
except:
MESSENGER.send_error("Error in reading the treefile.\n")
raise
tree_f.close()
if len(tree_list) > 1:
MESSENGER.send_warning(
'%d starting trees found in "%s". The first tree will be used.'
% (len(tree_list), tree_file))
starting_tree = tree_list[0]
score = None
tree_as_tmp_filename_to_report = tree_file
############################################################################
# This will relabel the taxa if they have problematic names
#####
multilocus_dataset.relabel_for_pasta()
############################################################################
# This ensures all nucleotide data is DNA internally
#####
restore_to_rna = False
if user_config.commandline.datatype.upper() == 'RNA':
multilocus_dataset.convert_rna_to_dna()
user_config.commandline.datatype = 'DNA'
restore_to_rna = True
export_names = True
if export_names:
try:
name_filename = pasta_products.get_abs_path_for_tag(
'name_translation.txt')
name_output = open(name_filename, 'w')
safe2real = multilocus_dataset.safe_to_real_names
safe_list = list(safe2real.keys())
safe_list.sort()
for safe in safe_list:
orig = safe2real[safe][0]
name_output.write("%s\n%s\n\n" % (safe, orig))
name_output.close()
MESSENGER.send_info(
"Name translation information saved to %s as safe name, original name, blank line format."
% name_filename)
except:
MESSENGER.send_info(
"Error exporting saving name translation to %s" %
name_filename)
if options.aligned:
options.aligned = all([i.is_aligned() for i in multilocus_dataset])
############################################################################
# Be prepared to kill any long running jobs
#####
prev_signals = []
for sig in [
signal.SIGTERM, signal.SIGABRT, signal.SIGINT
]: # signal.SIGABRT, signal.SIGBUS, signal.SIGINT, signal.SIGKILL, signal.SIGSTOP]:
prev_handler = signal.signal(sig, killed_handler)
prev_signals.append((sig, prev_handler))
try:
pasta_config_dict = pasta_config.dict()
if (not options.two_phase) and tree_file:
# getting the newick string here will allow us to get a string that is in terms of the correct taxon labels
starting_tree_str = str(starting_tree)
else:
if not options.two_phase:
MESSENGER.send_info(
"Creating a starting tree for the PASTA algorithm...")
if (options.two_phase) or (not options.aligned):
MESSENGER.send_info(
"Performing initial alignment of the entire data matrix..."
)
init_aln_dir = os.path.join(temporaries_dir, 'init_aln')
init_aln_dir = pasta_team.temp_fs.create_subdir(init_aln_dir)
delete_aln_temps = not (options.keeptemp
and options.keepalignmenttemps)
aln_job_list = []
query_fns = []
for unaligned_seqs in multilocus_dataset:
#backbone = sorted(unaligned_seqs.keys())[0:100]
backbone = sample(list(unaligned_seqs.keys()),
min(100, len(unaligned_seqs)))
backbone_seqs = unaligned_seqs.sub_alignment(backbone)
query_seq = list(
set(unaligned_seqs.keys()) - set(backbone))
qn = len(query_seq)
chunks = min(int(4 * pasta_config.num_cpus),
int(ceil(qn / 50.0)))
_LOG.debug(
"Will align the remaining %d sequences in %d chunks" %
(qn, chunks))
for ch in range(0, chunks):
query_fn = os.path.join(init_aln_dir,
"query-%d.fasta" % ch)
qa = unaligned_seqs.sub_alignment(
query_seq[ch:qn:chunks])
_LOG.debug("Chunk with %d sequences built" % len(qa))
qa.write_filepath(query_fn)
query_fns.append(query_fn)
job = pasta_team.aligner.create_job(
backbone_seqs,
tmp_dir_par=init_aln_dir,
context_str="initalign",
delete_temps=delete_aln_temps,
num_cpus=pasta_config.num_cpus)
aln_job_list.append(job)
_RunningJobs = aln_job_list
for job in aln_job_list:
jobq.put(job)
new_alignment = compact(job.get_results())
add_job_list = []
for query_fn in query_fns:
job = pasta_team.hmmeralign.create_job(
new_alignment,
query_fn,
tmp_dir_par=init_aln_dir,
context_str="initalign",
delete_temps=delete_aln_temps)
add_job_list.append(job)
_RunningJobs = None
for job in add_job_list:
jobq.put(job)
for job in add_job_list:
new_alignment.merge_in(compact(job.get_results()))
#new_alignment_list.apend(new_alignment)
#for locus_index, new_alignment in enumerate(new_alignment_list):
multilocus_dataset[0] = new_alignment
if delete_aln_temps:
pasta_team.temp_fs.remove_dir(init_aln_dir)
else:
MESSENGER.send_info(
"Input sequences assumed to be aligned (based on sequence lengths)."
)
MESSENGER.send_info(
"Performing initial tree search to get starting tree...")
init_tree_dir = os.path.join(temporaries_dir, 'init_tree')
init_tree_dir = pasta_team.temp_fs.create_subdir(init_tree_dir)
delete_tree_temps = not options.keeptemp
job = pasta_team.tree_estimator.create_job(
multilocus_dataset,
tmp_dir_par=init_tree_dir,
num_cpus=pasta_config.num_cpus,
context_str="inittree",
delete_temps=delete_tree_temps,
pasta_products=pasta_products,
step_num='initialsearch',
mask_gappy_sites=pasta_config_dict['mask_gappy_sites'])
_RunningJobs = job
jobq.put(job)
score, starting_tree_str = job.get_results()
_RunningJobs = None
alignment_as_tmp_filename_to_report = pasta_products.get_abs_path_for_iter_output(
"initialsearch", TEMP_SEQ_ALIGNMENT_TAG, allow_existing=True)
tree_as_tmp_filename_to_report = pasta_products.get_abs_path_for_iter_output(
"initialsearch", TEMP_TREE_TAG, allow_existing=True)
if delete_tree_temps:
pasta_team.temp_fs.remove_dir(init_tree_dir)
_LOG.debug('We have the tree and whole_alignment, partitions...')
if options.keeptemp:
pasta_config_dict['keep_iteration_temporaries'] = True
if options.keepalignmenttemps:
pasta_config_dict['keep_realignment_temporaries'] = True
job = PastaJob(multilocus_dataset=multilocus_dataset,
pasta_team=pasta_team,
name=options.job,
status_messages=MESSENGER.send_info,
score=score,
**pasta_config_dict)
if starting_tree is not None:
job.tree = generate_tree_with_splits_from_tree(
starting_tree, force_fully_resolved=True)
else:
job.tree_str = starting_tree_str
job.curr_iter_align_tmp_filename = alignment_as_tmp_filename_to_report
job.curr_iter_tree_tmp_filename = tree_as_tmp_filename_to_report
if score is not None:
job.store_optimum_results(
new_multilocus_dataset=multilocus_dataset,
new_tree_str=starting_tree_str,
new_score=score,
curr_timestamp=time.time())
if options.two_phase:
MESSENGER.send_info(
"Exiting with the initial tree because the PASTA algorithm is avoided when the --two-phase option is used."
)
else:
_RunningJobs = job
MESSENGER.send_info("Starting PASTA algorithm on initial tree...")
job.run(tmp_dir_par=temporaries_dir, pasta_products=pasta_products)
_RunningJobs = None
if job.return_final_tree_and_alignment:
alignment_as_tmp_filename_to_report = job.curr_iter_align_tmp_filename
else:
alignment_as_tmp_filename_to_report = job.best_alignment_tmp_filename
if user_config.commandline.raxml_search_after:
raxml_model = user_config.raxml.model.strip()
if not raxml_model:
dt = user_config.commandline.datatype
mf = pasta_team.tree_estimator.model
ms = fasttree_to_raxml_model_str(dt, mf)
pasta_team.raxml_tree_estimator.model = ms
rte = pasta_team.raxml_tree_estimator
MESSENGER.send_info(
"Performing post-processing tree search in RAxML...")
post_tree_dir = os.path.join(temporaries_dir, 'post_tree')
post_tree_dir = pasta_team.temp_fs.create_subdir(post_tree_dir)
delete_tree_temps = not options.keeptemp
starting_tree = None
if user_config.sate.start_tree_search_from_current:
starting_tree = job.tree
post_job = rte.create_job(
job.multilocus_dataset,
starting_tree=starting_tree,
num_cpus=pasta_config.num_cpus,
context_str="postraxtree",
tmp_dir_par=post_tree_dir,
delete_temps=delete_tree_temps,
pasta_products=pasta_products,
step_num="postraxtree",
mask_gappy_sites=pasta_config_dict['mask_gappy_sites'])
_RunningJobs = post_job
jobq.put(post_job)
post_score, post_tree = post_job.get_results()
_RunningJobs = None
tree_as_tmp_filename_to_report = pasta_products.get_abs_path_for_iter_output(
"postraxtree", TEMP_TREE_TAG, allow_existing=True)
if delete_tree_temps:
pasta_team.temp_fs.remove_dir(post_tree_dir)
job.tree_str = post_tree
job.score = post_score
if post_score > job.best_score:
job.best_tree_str = post_tree
job.best_score = post_score
else:
if job.return_final_tree_and_alignment:
tree_as_tmp_filename_to_report = job.curr_iter_tree_tmp_filename
else:
tree_as_tmp_filename_to_report = job.best_tree_tmp_filename
#######################################################################
# Restore original taxon names and RNA characters
#####
job.multilocus_dataset.restore_taxon_names()
if restore_to_rna:
job.multilocus_dataset.convert_dna_to_rna()
user_config.commandline.datatype = 'RNA'
assert len(pasta_products.alignment_streams) == len(
job.multilocus_dataset)
for i, alignment in enumerate(job.multilocus_dataset):
alignment_stream = pasta_products.alignment_streams[i]
MESSENGER.send_info("Writing resulting alignment to %s" %
alignment_stream.name)
alignment.write(alignment_stream, file_format="FASTA")
alignment_stream.close()
MESSENGER.send_info("Writing resulting tree to %s" %
pasta_products.tree_stream.name)
tree_str = job.tree.compose_newick()
pasta_products.tree_stream.write("%s;\n" % tree_str)
pasta_products.tree_stream.close()
#outtree_fn = options.result
#if outtree_fn is None:
# outtree_fn = os.path.join(seqdir, "combined_%s.tre" % options.job)
# else:
# outtree_fn = aln_filename + ".tre"
#MESSENGER.send_info("Writing resulting tree to %s" % outtree_fn)
#tree_str = str(job.tree)
#pasta_products.tree_stream.write("%s;\n" % tree_str)
MESSENGER.send_info("Writing resulting likelihood score to %s" %
pasta_products.score_stream.name)
pasta_products.score_stream.write("%s\n" % job.score)
pasta_products.score_stream.close()
if alignment_as_tmp_filename_to_report is not None:
MESSENGER.send_info(
'The resulting alignment (with the names in a "safe" form) was first written as the file "%s"'
% alignment_as_tmp_filename_to_report)
if tree_as_tmp_filename_to_report is not None:
MESSENGER.send_info(
'The resulting tree (with the names in a "safe" form) was first written as the file "%s"'
% tree_as_tmp_filename_to_report)
finally:
stop_worker()
for el in prev_signals:
sig, prev_handler = el
if prev_handler is None:
signal.signal(sig, signal.SIG_DFL)
else:
signal.signal(sig, prev_handler)
def finish_pasta_execution(pasta_team,
user_config,
temporaries_dir,
pasta_products,
multilocus_dataset):
global _RunningJobs
options = user_config.commandline
user_config.save_to_filepath(os.path.join(temporaries_dir, 'last_used.cfg'))
if options.timesfile:
f = open_with_intermediates(options.timesfile, 'a')
f.close()
set_timing_log_filepath(options.timesfile)
############################################################################
# Launch threads to do work
#####
pasta_config = user_config.get("sate")
start_worker(pasta_config.num_cpus)
#_LOG.debug("start reading the input alignment")
#multilocus_dataset = read_input_sequences(user_config.input_seq_filepaths,
# datatype=user_config.commandline.datatype,
# missing=user_config.commandline.missing)
############################################################################
# We must read the incoming tree in before we call the get_sequences_for_pasta
# function that relabels that taxa in the dataset
######
alignment_as_tmp_filename_to_report = None
tree_as_tmp_filename_to_report = None
starting_tree = None
tree_file = options.treefile
if tree_file:
if not os.path.exists(tree_file):
raise Exception('The tree file "%s" does not exist' % tree_file)
tree_f = open(tree_file, 'rU')
MESSENGER.send_info('Reading starting trees from "%s"...' % tree_file)
try:
tree_list = read_and_encode_splits(multilocus_dataset.dataset, tree_f,
starting_tree=True)
except KeyError:
MESSENGER.send_error("Error in reading the treefile, probably due to a name in the tree that does not match the names in the input sequence files.\n")
raise
except:
MESSENGER.send_error("Error in reading the treefile.\n")
raise
tree_f.close()
if len(tree_list) > 1:
MESSENGER.send_warning('%d starting trees found in "%s". The first tree will be used.' % (len(tree_list), tree_file))
starting_tree = tree_list[0]
score = None
tree_as_tmp_filename_to_report = tree_file
############################################################################
# This will relabel the taxa if they have problematic names
#####
multilocus_dataset.relabel_for_pasta()
############################################################################
# This ensures all nucleotide data is DNA internally
#####
restore_to_rna = False
if user_config.commandline.datatype.upper() == 'RNA':
multilocus_dataset.convert_rna_to_dna()
user_config.commandline.datatype = 'DNA'
restore_to_rna = True
export_names = True
if export_names:
try:
name_filename = pasta_products.get_abs_path_for_tag('name_translation.txt')
name_output = open(name_filename, 'w')
safe2real = multilocus_dataset.safe_to_real_names
safe_list = safe2real.keys()
safe_list.sort()
for safe in safe_list:
orig = safe2real[safe][0]
name_output.write("%s\n%s\n\n" % (safe, orig))
name_output.close()
MESSENGER.send_info("Name translation information saved to %s as safe name, original name, blank line format." % name_filename)
except:
MESSENGER.send_info("Error exporting saving name translation to %s" % name_filename)
if options.aligned:
options.aligned = all( [i.is_aligned() for i in multilocus_dataset] )
############################################################################
# Be prepared to kill any long running jobs
#####
prev_signals = []
for sig in [signal.SIGTERM, signal.SIGABRT, signal.SIGINT]: # signal.SIGABRT, signal.SIGBUS, signal.SIGINT, signal.SIGKILL, signal.SIGSTOP]:
prev_handler = signal.signal(sig, killed_handler)
prev_signals.append((sig, prev_handler))
try:
pasta_config_dict = pasta_config.dict()
if (not options.two_phase) and tree_file:
# getting the newick string here will allow us to get a string that is in terms of the correct taxon labels
starting_tree_str = starting_tree.compose_newick()
else:
if not options.two_phase:
MESSENGER.send_info("Creating a starting tree for the PASTA algorithm...")
if (options.two_phase) or (not options.aligned):
MESSENGER.send_info("Performing initial alignment of the entire data matrix...")
init_aln_dir = os.path.join(temporaries_dir, 'init_aln')
init_aln_dir = pasta_team.temp_fs.create_subdir(init_aln_dir)
delete_aln_temps = not (options.keeptemp and options.keepalignmenttemps)
aln_job_list = []
query_fns = []
for unaligned_seqs in multilocus_dataset:
#backbone = sorted(unaligned_seqs.keys())[0:100]
backbone = sample(unaligned_seqs.keys(), min(100,len(unaligned_seqs)))
backbone_seqs = unaligned_seqs.sub_alignment(backbone)
query_seq=list(set(unaligned_seqs.keys()) - set(backbone))
qn = len(query_seq)
chunks = min(int(4*pasta_config.num_cpus),int(ceil(qn/50.0)))
_LOG.debug("Will align the remaining %d sequences in %d chunks" %(qn,chunks))
for ch in xrange(0,chunks):
query_fn = os.path.join(init_aln_dir, "query-%d.fasta"%ch)
qa = unaligned_seqs.sub_alignment(query_seq[ch:qn:chunks])
_LOG.debug("Chunk with %d sequences built" %len(qa))
qa.write_filepath(query_fn)
query_fns.append(query_fn)
job = pasta_team.aligner.create_job(backbone_seqs,
tmp_dir_par=init_aln_dir,
context_str="initalign",
delete_temps=delete_aln_temps,
num_cpus=pasta_config.num_cpus)
aln_job_list.append(job)
_RunningJobs = aln_job_list
for job in aln_job_list:
jobq.put(job)
new_alignment = compact(job.get_results())
add_job_list = []
for query_fn in query_fns:
job = pasta_team.hmmeralign.create_job(new_alignment, query_fn,
tmp_dir_par=init_aln_dir,
context_str="initalign",
delete_temps=delete_aln_temps)
add_job_list.append(job)
_RunningJobs = None
for job in add_job_list:
jobq.put(job)
for job in add_job_list:
new_alignment.merge_in(compact(job.get_results()))
#new_alignment_list.apend(new_alignment)
#for locus_index, new_alignment in enumerate(new_alignment_list):
multilocus_dataset[0] = new_alignment
if delete_aln_temps:
pasta_team.temp_fs.remove_dir(init_aln_dir)
else:
MESSENGER.send_info("Input sequences assumed to be aligned (based on sequence lengths).")
MESSENGER.send_info("Performing initial tree search to get starting tree...")
init_tree_dir = os.path.join(temporaries_dir, 'init_tree')
init_tree_dir = pasta_team.temp_fs.create_subdir(init_tree_dir)
delete_tree_temps = not options.keeptemp
job = pasta_team.tree_estimator.create_job(multilocus_dataset,
tmp_dir_par=init_tree_dir,
num_cpus=pasta_config.num_cpus,
context_str="inittree",
delete_temps=delete_tree_temps,
pasta_products=pasta_products,
step_num='initialsearch',
mask_gappy_sites = pasta_config_dict['mask_gappy_sites'])
_RunningJobs = job
jobq.put(job)
score, starting_tree_str = job.get_results()
_RunningJobs = None
alignment_as_tmp_filename_to_report = pasta_products.get_abs_path_for_iter_output("initialsearch", TEMP_SEQ_ALIGNMENT_TAG, allow_existing=True)
tree_as_tmp_filename_to_report = pasta_products.get_abs_path_for_iter_output("initialsearch", TEMP_TREE_TAG, allow_existing=True)
if delete_tree_temps:
pasta_team.temp_fs.remove_dir(init_tree_dir)
_LOG.debug('We have the tree and whole_alignment, partitions...')
if options.keeptemp:
pasta_config_dict['keep_iteration_temporaries'] = True
if options.keepalignmenttemps:
pasta_config_dict['keep_realignment_temporaries'] = True
job = PastaJob(multilocus_dataset=multilocus_dataset,
pasta_team=pasta_team,
name=options.job,
status_messages=MESSENGER.send_info,
score=score,
**pasta_config_dict)
if starting_tree is not None:
job.tree = generate_tree_with_splits_from_tree(starting_tree, force_fully_resolved = True)
else:
job.tree_str = starting_tree_str
job.curr_iter_align_tmp_filename = alignment_as_tmp_filename_to_report
job.curr_iter_tree_tmp_filename = tree_as_tmp_filename_to_report
if score is not None:
job.store_optimum_results(new_multilocus_dataset=multilocus_dataset,
new_tree_str=starting_tree_str,
new_score=score,
curr_timestamp=time.time())
if options.two_phase:
MESSENGER.send_info("Exiting with the initial tree because the PASTA algorithm is avoided when the --two-phase option is used.")
else:
_RunningJobs = job
MESSENGER.send_info("Starting PASTA algorithm on initial tree...")
job.run(tmp_dir_par=temporaries_dir, pasta_products=pasta_products)
_RunningJobs = None
if job.return_final_tree_and_alignment:
alignment_as_tmp_filename_to_report = job.curr_iter_align_tmp_filename
else:
alignment_as_tmp_filename_to_report = job.best_alignment_tmp_filename
if user_config.commandline.raxml_search_after:
raxml_model = user_config.raxml.model.strip()
if not raxml_model:
dt = user_config.commandline.datatype
mf = pasta_team.tree_estimator.model
ms = fasttree_to_raxml_model_str(dt, mf)
pasta_team.raxml_tree_estimator.model = ms
rte = pasta_team.raxml_tree_estimator
MESSENGER.send_info("Performing post-processing tree search in RAxML...")
post_tree_dir = os.path.join(temporaries_dir, 'post_tree')
post_tree_dir = pasta_team.temp_fs.create_subdir(post_tree_dir)
delete_tree_temps = not options.keeptemp
starting_tree = None
if user_config.sate.start_tree_search_from_current:
starting_tree = job.tree
post_job = rte.create_job(job.multilocus_dataset,
starting_tree=starting_tree,
num_cpus=pasta_config.num_cpus,
context_str="postraxtree",
tmp_dir_par=post_tree_dir,
delete_temps=delete_tree_temps,
pasta_products=pasta_products,
step_num="postraxtree",
mask_gappy_sites = pasta_config_dict['mask_gappy_sites'])
_RunningJobs = post_job
jobq.put(post_job)
post_score, post_tree = post_job.get_results()
_RunningJobs = None
tree_as_tmp_filename_to_report = pasta_products.get_abs_path_for_iter_output("postraxtree", TEMP_TREE_TAG, allow_existing=True)
if delete_tree_temps:
pasta_team.temp_fs.remove_dir(post_tree_dir)
job.tree_str = post_tree
job.score = post_score
if post_score > job.best_score:
job.best_tree_str = post_tree
job.best_score = post_score
else:
if job.return_final_tree_and_alignment:
tree_as_tmp_filename_to_report = job.curr_iter_tree_tmp_filename
else:
tree_as_tmp_filename_to_report = job.best_tree_tmp_filename
#######################################################################
# Restore original taxon names and RNA characters
#####
job.multilocus_dataset.restore_taxon_names()
if restore_to_rna:
job.multilocus_dataset.convert_dna_to_rna()
user_config.commandline.datatype = 'RNA'
assert len(pasta_products.alignment_streams) == len(job.multilocus_dataset)
for i, alignment in enumerate(job.multilocus_dataset):
alignment_stream = pasta_products.alignment_streams[i]
MESSENGER.send_info("Writing resulting alignment to %s" % alignment_stream.name)
alignment.write(alignment_stream, file_format="FASTA")
alignment_stream.close()
MESSENGER.send_info("Writing resulting tree to %s" % pasta_products.tree_stream.name)
tree_str = job.tree.compose_newick()
pasta_products.tree_stream.write("%s;\n" % tree_str)
#outtree_fn = options.result
#if outtree_fn is None:
# if options.multilocus:
# outtree_fn = os.path.join(seqdir, "combined_%s.tre" % options.job)
# else:
# outtree_fn = aln_filename + ".tre"
#MESSENGER.send_info("Writing resulting tree to %s" % outtree_fn)
#tree_str = job.tree.compose_newick()
#pasta_products.tree_stream.write("%s;\n" % tree_str)
MESSENGER.send_info("Writing resulting likelihood score to %s" % pasta_products.score_stream.name)
pasta_products.score_stream.write("%s\n" % job.score)
if alignment_as_tmp_filename_to_report is not None:
MESSENGER.send_info('The resulting alignment (with the names in a "safe" form) was first written as the file "%s"' % alignment_as_tmp_filename_to_report)
if tree_as_tmp_filename_to_report is not None:
MESSENGER.send_info('The resulting tree (with the names in a "safe" form) was first written as the file "%s"' % tree_as_tmp_filename_to_report)
finally:
stop_worker()
for el in prev_signals:
sig, prev_handler = el
if prev_handler is None:
signal.signal(sig, signal.SIG_DFL)
else:
signal.signal(sig, prev_handler)
