def run_test(self, options):
"""Run test of classify workflow."""
make_sure_path_exists(options.out_dir)
output_dir = os.path.join(options.out_dir, 'output')
genome_test_dir = os.path.join(options.out_dir, 'genomes')
if os.path.exists(genome_test_dir):
self.logger.error(
'Test directory {} already exists. Test must be run with a new directory.'
.format(genome_test_dir))
sys.exit(-1)
current_path = os.path.dirname(os.path.realpath(__file__))
input_dir = os.path.join(current_path, 'tests', 'data', 'genomes')
shutil.copytree(input_dir, genome_test_dir)
cmd = 'gtdbtk classify_wf --genome_dir {} --out_dir {} --cpus {}'.format(
genome_test_dir, output_dir, options.cpus)
print("Command:")
print(cmd)
os.system(cmd)
summary_file = os.path.join(
output_dir, PATH_AR122_SUMMARY_OUT.format(prefix='gtdbtk'))
if not os.path.exists(summary_file):
print("{} is missing.\nTest has failed.".format(summary_file))
sys.exit(-1)
self.logger.info('Test has successfully finished.')
def run_test(self, options):
make_sure_path_exists(options.out_dir)
genome_test_dir = os.path.join(options.out_dir, 'genomes')
output_dir = os.path.join(options.out_dir, 'output')
if os.path.isdir(genome_test_dir):
shutil.rmtree(genome_test_dir)
current_path = os.path.dirname(os.path.realpath(__file__))
input_dir = os.path.join(current_path, 'tests', 'data', 'genomes')
shutil.copytree(input_dir, genome_test_dir)
cmd = 'gtdbtk classify_wf --genome_dir {} --out_dir {} --cpus {}'.format(
genome_test_dir, output_dir, options.cpus)
print "Command:"
print cmd
os.system(cmd)
summary_file = os.path.join(output_dir, 'gtdbtk.ar122.summary.tsv')
if not os.path.exists(summary_file):
print "{} is missing.\nTest has failed.".format(summary_file)
sys.exit(-1)
self.logger.info('Test has successfully finished.')
def export_msa(self, domain, output_file):
file_to_export = Config.CONCAT_BAC120
if domain == 'arc':
file_to_export = Config.CONCAT_AR122
make_sure_path_exists(os.path.dirname(output_file))
copyfile(file_to_export, output_file)
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 classify(self, options):
"""Determine taxonomic classification of genomes."""
check_dir_exists(options.align_dir)
make_sure_path_exists(options.out_dir)
if options.scratch_dir:
make_sure_path_exists(options.scratch_dir)
genomes = self._genomes_to_process(options.genome_dir,
options.batchfile,
options.extension)
classify = Classify(options.cpus)
classify.run(genomes, options.align_dir, options.out_dir,
options.prefix, options.scratch_dir, options.debug)
self.logger.info('Done.')
def align(self, options):
"""Create MSA from marker genes."""
check_dir_exists(options.identify_dir)
make_sure_path_exists(options.out_dir)
if not hasattr(options, 'outgroup_taxon'):
options.outgroup_taxon = None
markers = Markers(options.cpus)
markers.align(options.identify_dir, options.skip_gtdb_refs,
options.taxa_filter, options.min_perc_aa,
options.custom_msa_filters, options.rnd_seed,
options.cols_per_gene, options.min_consensus,
options.max_consensus, options.min_perc_taxa,
options.out_dir, options.prefix, options.outgroup_taxon)
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.')