def decorate(self, options):
"""Decorate tree with GTDB taxonomy."""
check_file_exists(options.input_tree)
# Config.TAXONOMY_FILE
self.logger.warning('NOT YET IMPLEMENTED!')
self.logger.info('Done.')
def infer(self, options):
"""Infer tree from MSA."""
check_file_exists(options.msa_file)
make_sure_path_exists(options.out_dir)
if options.cpus > 1:
check_dependencies(['FastTreeMP'])
else:
check_dependencies(['FastTree'])
self.logger.info('Inferring tree with FastTree using %s+GAMMA.' %
options.prot_model)
if hasattr(options, 'suffix'):
output_tree = os.path.join(
options.out_dir,
options.prefix + options.suffix + '.unrooted.tree')
tree_log = os.path.join(
options.out_dir, options.prefix + options.suffix + '.tree.log')
fasttree_log = os.path.join(
options.out_dir,
options.prefix + options.suffix + '.fasttree.log')
else:
output_tree = os.path.join(options.out_dir,
options.prefix + '.unrooted.tree')
tree_log = os.path.join(options.out_dir,
options.prefix + '.tree.log')
fasttree_log = os.path.join(options.out_dir,
options.prefix + '.fasttree.log')
if options.prot_model == 'JTT':
model_str = ''
elif options.prot_model == 'WAG':
model_str = ' -wag'
elif options.prot_model == 'LG':
model_str = ' -lg'
support_str = ''
if options.no_support:
support_str = ' -nosupport'
gamma_str = ' -gamma'
if options.no_gamma:
gamma_str = ''
cmd = '-quiet%s%s%s -log %s %s > %s 2> %s' % (
support_str, model_str, gamma_str, tree_log, options.msa_file,
output_tree, fasttree_log)
if options.cpus > 1:
cmd = 'FastTreeMP ' + cmd
else:
cmd = 'FastTree ' + cmd
self.logger.info('Running: %s' % cmd)
os.system(cmd)
self.logger.info('Done.')
def root(self, options):
"""Root tree using outgroup."""
self.logger.warning("Tree rooting is still under development!")
check_file_exists(options.input_tree)
gtdb_taxonomy = Taxonomy().read(Config.TAXONOMY_FILE)
self.logger.info('Identifying genomes from the specified outgroup.')
outgroup = set()
for genome_id, taxa in gtdb_taxonomy.iteritems():
if options.outgroup_taxon in taxa:
outgroup.add(genome_id)
reroot = RerootTree()
reroot.root_with_outgroup(options.input_tree, options.output_tree,
outgroup)
self.logger.info('Done.')
def identify(self, options):
"""Identify marker genes in genomes."""
if options.genome_dir:
check_dir_exists(options.genome_dir)
if options.batchfile:
check_file_exists(options.batchfile)
make_sure_path_exists(options.out_dir)
genomes = self._genomes_to_process(options.genome_dir,
options.batchfile,
options.extension)
markers = Markers(options.cpus)
markers.identify(genomes, options.out_dir, options.prefix)
self.logger.info('Done.')
def infer(self, options):
"""Infer tree from MSA."""
check_file_exists(options.msa_file)
make_sure_path_exists(options.out_dir)
if options.cpus > 1:
check_dependencies(['FastTreeMP'])
os.environ['OMP_NUM_THREADS'] = '%d' % options.cpus
else:
check_dependencies(['FastTree'])
self.logger.info('Inferring tree with FastTree using %s+GAMMA.' %
options.prot_model)
if hasattr(options, 'suffix'):
output_tree = os.path.join(
options.out_dir,
PATH_MARKER_UNROOTED_TREE.format(prefix=options.prefix,
marker=options.suffix))
tree_log = os.path.join(
options.out_dir,
PATH_MARKER_TREE_LOG.format(prefix=options.prefix,
marker=options.suffix))
fasttree_log = os.path.join(
options.out_dir,
PATH_MARKER_FASTTREE_LOG.format(prefix=options.prefix,
marker=options.suffix))
else:
output_tree = os.path.join(
options.out_dir,
PATH_UNROOTED_TREE.format(prefix=options.prefix))
tree_log = os.path.join(
options.out_dir, PATH_TREE_LOG.format(prefix=options.prefix))
fasttree_log = os.path.join(
options.out_dir,
PATH_FASTTREE_LOG.format(prefix=options.prefix))
make_sure_path_exists(os.path.dirname(output_tree))
make_sure_path_exists(os.path.dirname(tree_log))
make_sure_path_exists(os.path.dirname(fasttree_log))
if options.prot_model == 'JTT':
model_str = ''
elif options.prot_model == 'WAG':
model_str = ' -wag'
elif options.prot_model == 'LG':
model_str = ' -lg'
support_str = ''
if options.no_support:
support_str = ' -nosupport'
gamma_str = ' -gamma'
gamma_str_info = '+GAMMA'
if options.no_gamma:
gamma_str = ''
gamma_str_info = ''
self.logger.info('Inferring tree with FastTree using {}.'.format(
options.prot_model, gamma_str_info))
cmd = '-quiet%s%s%s -log %s %s > %s 2> %s' % (
support_str, model_str, gamma_str, tree_log, options.msa_file,
output_tree, fasttree_log)
if options.cpus > 1:
cmd = 'FastTreeMP ' + cmd
else:
cmd = 'FastTree ' + cmd
self.logger.info('Running: %s' % cmd)
os.system(cmd)
self.logger.info('Done.')