def identify(self, options):
"""Identify marker genes in genomes.
Parameters
----------
options : argparse.Namespace
The CLI arguments input by the user.
"""
if options.genome_dir:
assert_dir_exists(options.genome_dir)
if options.batchfile:
assert_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,
options.force, options.genes)
self.logger.info('Done.')
def infer(self, options):
"""Infer a tree from a user specified MSA.
Parameters
----------
options : argparse.Namespace
The CLI arguments input by the user.
"""
assert_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'])
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))
fasttree = FastTree()
fasttree.run(output_tree, tree_log, fasttree_log, options.prot_model,
options.no_support, options.no_gamma, options.msa_file,
options.cpus)
if hasattr(options,
'subparser_name') and options.subparser_name == 'infer':
symlink_f(
output_tree[len(options.out_dir) + 1:],
os.path.join(options.out_dir, os.path.basename(output_tree)))
self.logger.info('Done.')
def decorate(self, options):
"""Decorate tree with GTDB taxonomy.
Parameters
----------
options : argparse.Namespace
The CLI arguments input by the user.
"""
assert_file_exists(options.input_tree)
# Config.TAXONOMY_FILE
self.logger.warning('DECORATE NOT YET IMPLEMENTED!')
self.logger.info('Done.')
def root(self, options):
"""Root tree using outgroup.
Parameters
----------
options : argparse.Namespace
The CLI arguments input by the user.
"""
self.logger.warning("Tree rooting is still under development!")
assert_file_exists(options.input_tree)
if options.custom_taxonomy_file:
assert_file_exists(options.custom_taxonomy_file)
taxonomy = Taxonomy().read(options.custom_taxonomy_file)
else:
taxonomy = Taxonomy().read(Config.TAXONOMY_FILE)
self.logger.info('Identifying genomes from the specified outgroup.')
outgroup = set()
for genome_id, taxa in 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)
# Symlink to the tree summary file, if not run independently
if hasattr(options, 'suffix'):
if options.suffix == 'bac120':
symlink_f(
PATH_BAC120_ROOTED_TREE.format(prefix=options.prefix),
os.path.join(
options.out_dir,
os.path.basename(
PATH_AR122_ROOTED_TREE.format(
prefix=options.prefix))))
elif options.suffix == 'ar122':
symlink_f(
PATH_AR122_ROOTED_TREE.format(prefix=options.prefix),
os.path.join(
options.out_dir,
os.path.basename(
PATH_AR122_ROOTED_TREE.format(
prefix=options.prefix))))
else:
raise GenomeMarkerSetUnknown(
'There was an error determining the marker set.')
self.logger.info('Done.')
