def test_classify(self):
tmp_folder = ''.join(
random.choice(string.ascii_uppercase + string.digits)
for _ in range(10))
classify_options = self.options
classify_options.genome_dir = self.genome_dir
classify_options.align_dir = self.align_dir_reference
classify_options.out_dir = os.path.join(self.generic_out_path,
tmp_folder, 'classify')
classify_options.recalculate_red = False
classify_options.split_tree = False
self.optionparser.classify(classify_options)
summary_fh = ClassifySummaryFileAR122(classify_options.out_dir,
classify_options.prefix)
summary_fh.read()
self.assertEqual(
'd__Archaea;p__Methanobacteriota;c__Methanobacteria;o__Methanobacteriales;f__Methanobacteriaceae;g__Methanobrevibacter;s__Methanobrevibacter ruminantium',
summary_fh.rows['genome_1'].classification)
self.assertEqual(
'd__Archaea;p__Thermoplasmatota;c__Thermoplasmata;o__Methanomassiliicoccales;f__Methanomethylophilaceae;g__VadinCA11;s__VadinCA11 sp002498365',
summary_fh.rows['genome_2'].classification)
self.assertEqual(
'd__Archaea;p__Thermoplasmatota;c__Thermoplasmata;o__Methanomassiliicoccales;f__Methanomethylophilaceae;g__VadinCA11;s__VadinCA11 sp002498365',
summary_fh.rows['genome_3'].classification)
def run_test(self, options):
"""Run test of classify workflow.
Parameters
----------
options : argparse.Namespace
The CLI arguments input by the user.
Returns
-------
bool
True if the test succeeds.
Raises
------
GTDBTkTestFailure
If the test fails.
"""
# Use a temporary directory if none is supplied.
if options.out_dir:
out_dir_fh = None
make_sure_path_exists(options.out_dir)
else:
out_dir_fh = tempfile.TemporaryDirectory(prefix='gtdbtk_tmp_')
options.out_dir = out_dir_fh.name
self.logger.info('Using a temporary directory as out_dir was not specified.')
try:
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(f'Test directory {genome_test_dir} already exists.')
self.logger.error('Test must be run in a new directory.')
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)
args = ['gtdbtk', 'classify_wf', '--genome_dir', genome_test_dir,
'--out_dir', output_dir, '--cpus', str(options.cpus)]
self.logger.info('Command: {}'.format(' '.join(args)))
# Pipe the output and write to disk.
path_stdout = os.path.join(options.out_dir, 'test_execution.log')
with subprocess.Popen(args, stdout=subprocess.PIPE,
stderr=subprocess.PIPE, encoding='utf-8') as proc:
with open(path_stdout, 'w') as fh_stdout:
bar_fmt = ' <TEST OUTPUT> '.center(22) + '{desc}'
with tqdm(bar_format=bar_fmt, leave=False) as p_bar:
while True:
line = proc.stdout.readline()
if not line:
break
fh_stdout.write(f'{line}')
p_bar.set_description_str(line.strip())
proc.wait()
exit_code = proc.returncode
summary_fh = ClassifySummaryFileAR122(output_dir, 'gtdbtk')
if exit_code != 0:
self.logger.error('The test returned a non-zero exit code.')
self.logger.error('A detailed summary of the execution log can be '
'found here: {}'.format(path_stdout))
self.logger.error('The test has failed.')
sys.exit(1)
if not os.path.exists(summary_fh.path):
self.logger.error(f"{summary_fh.path} is missing.")
self.logger.error('A detailed summary of the execution log can be '
'found here: {}'.format(path_stdout))
self.logger.error('The test has failed.')
sys.exit(1)
finally:
if out_dir_fh:
out_dir_fh.cleanup()
self.logger.info('Test has successfully finished.')
return True