def test_out_dir(self):
cfg = Config(self.defaults, self.parsed_args)
self.assertEqual(
cfg.out_dir(),
os.path.join(cfg.res_search_dir(),
f'macsyfinder-{strftime("%Y%m%d_%H-%M-%S")}'))
self.parsed_args.out_dir = 'foo'
cfg = Config(self.defaults, self.parsed_args)
self.assertEqual(cfg.out_dir(), 'foo')
def test_Config(self):
cfg = Config(self.defaults, self.parsed_args)
methods_needing_args = {
'inter_gene_max_space': None,
'max_nb_genes': None,
'min_genes_required': None,
'min_mandatory_genes_required': None,
'multi_loci': None
}
for opt, val in self.defaults.items():
if opt == 'out_dir':
self.assertEqual(
cfg.out_dir(),
os.path.join(cfg.res_search_dir(),
f'macsyfinder-{strftime("%Y%m%d_%H-%M-%S")}'))
elif opt == 'multi_loci':
self.assertFalse(cfg.multi_loci('whatever'))
elif opt in methods_needing_args:
self.assertEqual(
getattr(cfg, opt)('whatever'),
val,
msg=
f"test of '{opt}' failed : expected{getattr(cfg, opt)('whatever')} != got {val}"
)
else:
self.assertEqual(
getattr(cfg, opt)(),
val,
msg=
f"test of '{opt}' failed : expected{getattr(cfg, opt)()} != got {val}"
)
def test_Config_file(self):
methods_needing_args = {
'inter_gene_max_space': [('set_1/Flagellum', 4),
('set_1/T2SS', 2)],
'max_nb_genes': [('set_1/Flagellum', 6), ('set_1/T3SS', 3)],
'min_genes_required': [('set_1/Flagellum', 8), ('set_1/T4SS', 4)],
'min_mandatory_genes_required': [('set_1/Flagellum', 12),
('set_1/T6SS', 6)],
'multi_loci': {'set_1/Flagellum', 'T4SS'}
}
self.parsed_args.cfg_file = self.find_data(
os.path.join('conf_files', 'macsy_models.conf'))
cfg = Config(self.defaults, self.parsed_args)
expected_values = {k: v for k, v in self.defaults.items()}
expected_values['cfg_file'] = self.parsed_args.cfg_file
expected_values.update(methods_needing_args)
for opt, val in expected_values.items():
if opt == 'out_dir':
self.assertEqual(
cfg.out_dir(),
os.path.join(cfg.res_search_dir(),
f'macsyfinder-{strftime("%Y%m%d_%H-%M-%S")}'))
elif opt == 'multi_loci':
self.assertTrue(cfg.multi_loci('set_1/Flagellum'))
self.assertTrue(cfg.multi_loci('set_1/T4SS'))
self.assertFalse(cfg.multi_loci('set_1/T6SS'))
elif opt in methods_needing_args:
for model, genes in expected_values[opt]:
self.assertEqual(getattr(cfg, opt)(model), genes)
else:
self.assertEqual(getattr(cfg, opt)(), val)
def test_Config_args(self):
methods_needing_args = {
'inter_gene_max_space': [('set_1/Flagellum', '14'),
('set_1/T2SS', '12')],
'max_nb_genes': [('set_1/Flagellum', '16'), ('set_1/T3SS', '13')],
'min_genes_required': [('set_1/Flagellum', '18'),
('set_1/T4SS', '14')],
'min_mandatory_genes_required': [('set_1/Flagellum', '22'),
('set_1/T6SS', '16')],
'multi_loci':
'set_1/Flagellum, set_1/T4SS',
}
for opt, value in methods_needing_args.items():
setattr(self.parsed_args, opt, value)
simple_opt = {
'hmmer': 'foo',
'i_evalue_sel': 20,
'replicon_topology': 'linear',
'db_type': 'gembase',
'sequence_db': self.find_data(os.path.join('base',
'test_1.fasta')),
'topology_file': __file__ # test only the existence of a file
}
for opt, val in simple_opt.items():
setattr(self.parsed_args, opt, val)
cfg = Config(self.defaults, self.parsed_args)
expected_values = {k: v for k, v in self.defaults.items()}
expected_values.update(methods_needing_args)
expected_values.update(simple_opt)
for opt, val in expected_values.items():
if opt == 'out_dir':
self.assertEqual(
cfg.out_dir(),
os.path.join(cfg.res_search_dir(),
f'macsyfinder-{strftime("%Y%m%d_%H-%M-%S")}'))
elif opt == 'multi_loci':
self.assertTrue(cfg.multi_loci('set_1/Flagellum'))
self.assertTrue(cfg.multi_loci('set_1/T4SS'))
self.assertFalse(cfg.multi_loci('set_1/T6SS'))
elif opt in methods_needing_args:
for model, genes in expected_values[opt]:
self.assertEqual(getattr(cfg, opt)(model), int(genes))
elif opt == 'models_dir':
self.assertEqual(
getattr(cfg, opt)(), self.defaults['system_models_dir'])
else:
self.assertEqual(
getattr(cfg, opt)(),
val,
msg=
f"{opt} failed: expected: val '{val}' != got '{getattr(cfg, opt)()}'"
)
def test_Config_file_n_args(self):
cfg_needing_args = {
'inter_gene_max_space': [('set_1/Flagellum', '4'),
('set_1/T2SS', '2')],
'max_nb_genes': [('set_1/Flagellum', '6'), ('set_1/T3SS', '3')],
'min_genes_required': [('set_1/Flagellum', '8'),
('set_1/T4SS', '4')],
'min_mandatory_genes_required': [('set_1/Flagellum', '12'),
('set_1/T6SS', '6')],
'multi_loci':
'set_1/Flagellum, set_1/T4SS',
}
self.parsed_args.cfg_file = self.find_data(
os.path.join('conf_files', 'macsy_models.conf'))
expected_values = {k: v for k, v in self.defaults.items()}
expected_values['cfg_file'] = self.parsed_args.cfg_file
expected_values.update(cfg_needing_args)
cmd_needing_args = {
'min_genes_required': [('set_1/Flagellum', 18), ('T4SS', 14)],
'min_mandatory_genes_required': [('set_1/Flagellum', 22),
('set_1/T6SS', 16)],
}
for opt, value in cmd_needing_args.items():
setattr(self.parsed_args, opt,
' '.join([f"{m} {v}" for m, v in value]))
simple_opt = {'hmmer': 'foo', 'i_evalue_sel': 20, 'db_type': 'gembase'}
for opt, val in simple_opt.items():
setattr(self.parsed_args, opt, val)
cfg = Config(self.defaults, self.parsed_args)
expected_values.update(cmd_needing_args)
expected_values.update(simple_opt)
for opt, exp_val in expected_values.items():
if opt == 'out_dir':
self.assertEqual(
cfg.out_dir(),
os.path.join(cfg.res_search_dir(),
f"macsyfinder-{strftime('%Y%m%d_%H-%M-%S')}"))
elif opt == 'multi_loci':
self.assertTrue(cfg.multi_loci('set_1/Flagellum'))
self.assertTrue(cfg.multi_loci('set_1/T4SS'))
self.assertFalse(cfg.multi_loci('set_1/T6SS'))
elif opt in cfg_needing_args:
for model, val in expected_values[opt]:
self.assertEqual(getattr(cfg, opt)(model), int(val))
elif opt == 'models_dir':
self.assertEqual(
getattr(cfg, opt)(), self.defaults['system_models_dir'])
else:
self.assertEqual(getattr(cfg, opt)(), exp_val)
def test_Config_conf_file_virtualenv(self):
methods_needing_args = {
'inter_gene_max_space': [('set_1/Flagellum', 4),
('set_1/T2SS', 2)],
'min_mandatory_genes_required': [('set_1/Flagellum', 9),
('set_1/T2SS', 5)]
}
modified_args = {
'db_type': 'gembase',
'coverage_profile': 0.1,
'replicon_topology': 'circular'
}
with tempfile.TemporaryDirectory() as tmpdirname:
ori_conf_file = self.find_data(
os.path.join('conf_files', 'macsy_virtualenv_test.conf'))
conf_dir = os.path.join(tmpdirname, 'etc', 'macsyfinder')
os.makedirs(conf_dir)
dest_conf_file = os.path.join(conf_dir, 'macsyfinder.conf')
shutil.copy(ori_conf_file, dest_conf_file)
virtual_env = os.environ.get("VIRTUAL_ENV")
os.environ['VIRTUAL_ENV'] = tmpdirname
try:
cfg = Config(self.defaults, self.parsed_args)
expected_values = {k: v for k, v in self.defaults.items()}
expected_values.update(methods_needing_args)
expected_values.update(modified_args)
for opt, val in expected_values.items():
if opt == 'out_dir':
self.assertEqual(
cfg.out_dir(),
os.path.join(
cfg.res_search_dir(),
f'macsyfinder-{strftime("%Y%m%d_%H-%M-%S")}'))
elif opt in ('max_nb_genes', 'min_genes_required',
'multi_loci'): # not set in cfg file
pass
elif opt in methods_needing_args:
for model, genes in expected_values[opt]:
self.assertEqual(getattr(cfg, opt)(model), genes)
elif opt == 'models_dir':
self.assertEqual(
getattr(cfg, opt)(),
self.defaults['system_models_dir'])
else:
self.assertEqual(getattr(cfg, opt)(), val)
finally:
if virtual_env:
os.environ["VIRTUAL_ENV"] = virtual_env
def test_Config_default_conf_file(self):
methods_needing_args = {
'inter_gene_max_space': [('set_1/Flagellum', 4),
('set_1/T2SS', 2)],
'max_nb_genes': [('set_1/Flagellum', 6), ('set_1/T3SS', 3)],
'min_genes_required': [('set_1/Flagellum', 8), ('set_1/T4SS', 4)],
'min_mandatory_genes_required': [('set_1/Flagellum', 12),
('set_1/T6SS', 6)],
'multi_loci': {'set_1/Flagellum', 'set_1/T4SS'}
}
with tempfile.TemporaryDirectory() as tmpdirname:
ori_conf_file = self.find_data(
os.path.join('conf_files', 'macsy_models.conf'))
dest_conf_file = os.path.join(tmpdirname, 'macsyfinder.conf')
shutil.copy(ori_conf_file, dest_conf_file)
os.environ['MACSY_CONF'] = dest_conf_file
virtual_env = os.environ.get("VIRTUAL_ENV")
del os.environ["VIRTUAL_ENV"]
try:
cfg = Config(self.defaults, self.parsed_args)
expected_values = {k: v for k, v in self.defaults.items()}
expected_values.update(methods_needing_args)
for opt, val in expected_values.items():
if opt == 'out_dir':
self.assertEqual(
cfg.out_dir(),
os.path.join(
cfg.res_search_dir(),
f'macsyfinder-{strftime("%Y%m%d_%H-%M-%S")}'))
elif opt == 'multi_loci':
self.assertTrue(cfg.multi_loci('set_1/Flagellum'))
self.assertTrue(cfg.multi_loci('set_1/T4SS'))
self.assertFalse(cfg.multi_loci('set_1/T6SS'))
elif opt in methods_needing_args:
for model, genes in expected_values[opt]:
self.assertEqual(getattr(cfg, opt)(model), genes)
elif opt == 'models_dir':
self.assertEqual(
getattr(cfg, opt)(),
self.defaults['system_models_dir'])
else:
self.assertEqual(getattr(cfg, opt)(), val)
finally:
os.environ["VIRTUAL_ENV"] = virtual_env
def test_project_config_file(self):
os.chdir(self.tmp_dir)
methods_needing_args = {
'inter_gene_max_space': [('set_1/Flagellum', 4),
('set_1/T2SS', 2)],
'max_nb_genes': [('set_1/Flagellum', 6), ('set_1/T3SS', 3)],
'min_genes_required': [('set_1/Flagellum', 8), ('set_1/T4SS', 4)],
'min_mandatory_genes_required': [('set_1/Flagellum', 12),
('set_1/T6SS', 6)],
'multi_loci': {'set_1/Flagellum', 'T4SS'}
}
hmmer_opts_in_file = {'coverage_profile': 0.8, 'e_value_search': 0.12}
try:
shutil.copyfile(
self.find_data(os.path.join('conf_files', 'project.conf')),
os.path.join(self.tmp_dir, 'macsyfinder.conf'))
cfg = Config(self.defaults, self.parsed_args)
expected_values = {k: v for k, v in self.defaults.items()}
expected_values.update(methods_needing_args)
expected_values.update(hmmer_opts_in_file)
for opt, val in expected_values.items():
if opt == 'out_dir':
self.assertEqual(
cfg.out_dir(),
os.path.join(
cfg.res_search_dir(),
f'macsyfinder-{strftime("%Y%m%d_%H-%M-%S")}'))
elif opt == 'multi_loci':
self.assertTrue(cfg.multi_loci('set_1/Flagellum'))
self.assertTrue(cfg.multi_loci('set_1/T4SS'))
self.assertFalse(cfg.multi_loci('set_1/T6SS'))
elif opt in methods_needing_args:
for model, genes in expected_values[opt]:
self.assertEqual(getattr(cfg, opt)(model), genes)
elif opt == 'models_dir':
self.assertEqual(
getattr(cfg, opt)(),
self.defaults['system_models_dir'])
else:
self.assertEqual(getattr(cfg, opt)(), val)
except:
os.chdir(self._current_dir)
class TestReport(MacsyTest):
def setUp(self):
args = argparse.Namespace()
args.db_type = 'gembase'
args.models_dir = self.find_data('models')
args.res_search_dir = tempfile.gettempdir()
args.log_level = 30
args.out_dir = os.path.join(args.res_search_dir,
'test_macsyfinder_Report')
if os.path.exists(args.out_dir):
shutil.rmtree(args.out_dir)
os.mkdir(args.out_dir)
seq_db = self.find_data("base", "test_base.fa")
shutil.copy(seq_db, args.out_dir)
args.sequence_db = os.path.join(args.out_dir, os.path.basename(seq_db))
self.cfg = Config(MacsyDefaults(), args)
os.mkdir(os.path.join(self.cfg.out_dir(), self.cfg.hmmer_dir()))
self.model_name = 'foo'
self.model_location = ModelLocation(
path=os.path.join(args.models_dir, self.model_name))
# we need to reset the ProfileFactory
# because it's a like a singleton
# so other tests are influenced by ProfileFactory and it's configuration
# for instance search_genes get profile without hmmer_exe
self.profile_factory = ProfileFactory(self.cfg)
idx = Indexes(self.cfg)
idx.build()
def tearDown(self):
try:
shutil.rmtree(self.cfg.working_dir())
except Exception:
pass
def test_working_dir(self):
cfg = Config(self.defaults, self.parsed_args)
self.assertEqual(cfg.out_dir(), cfg.working_dir())
class TestSearchGenes(MacsyTest):
def setUp(self):
self.tmp_dir = os.path.join(tempfile.gettempdir(),
'test_macsyfinder_search_genes')
if os.path.exists(self.tmp_dir):
shutil.rmtree(self.tmp_dir)
os.mkdir(self.tmp_dir)
args = argparse.Namespace()
args.sequence_db = self.find_data("base", "test_base.fa")
args.db_type = 'gembase'
args.models_dir = self.find_data('models')
args.log_level = 30
args.out_dir = os.path.join(self.tmp_dir, 'job_1')
args.res_search_dir = args.out_dir
os.mkdir(args.out_dir)
self.cfg = Config(MacsyDefaults(), args)
self.model_name = 'foo'
self.model_location = ModelLocation(
path=os.path.join(args.models_dir, self.model_name))
idx = Indexes(self.cfg)
idx._build_my_indexes()
self.profile_factory = ProfileFactory(self.cfg)
def tearDown(self):
try:
shutil.rmtree(self.cfg.working_dir())
#pass
except:
pass
@unittest.skipIf(not which('hmmsearch'), 'hmmsearch not found in PATH')
def test_search(self):
gene_name = "abc"
c_gene_abc = CoreGene(self.model_location, gene_name,
self.profile_factory)
report = search_genes([c_gene_abc], self.cfg)
expected_hit = [
Hit(c_gene_abc, "ESCO030p01_000260", 706, "ESCO030p01", 26,
float(1.000e-200), float(660.800), float(1.000), float(0.714),
160, 663)
]
self.assertEqual(len(report), 1)
self.assertEqual(expected_hit[0], report[0].hits[0])
@unittest.skipIf(not which('hmmsearch'), 'hmmsearch not found in PATH')
def test_search_recover(self):
# first job searching using hmmsearch
gene_name = "abc"
c_gene_abc = CoreGene(self.model_location, gene_name,
self.profile_factory)
report = search_genes([c_gene_abc], self.cfg)
expected_hit = [
Hit(c_gene_abc, "ESCO030p01_000260", 706, "ESCO030p01", 26,
float(1.000e-200), float(660.800), float(1.000), float(0.714),
160, 663)
]
# second job using recover
# disable hmmer to be sure that test use the recover inner function
self.cfg.hmmer = lambda: "hmmer_disable"
# and create a new dir for the second job
previous_job_path = self.cfg.working_dir()
self.cfg.previous_run = lambda: previous_job_path
self.cfg.out_dir = lambda: os.path.join(self.tmp_dir, 'job_2')
os.mkdir(self.cfg.out_dir())
# rerun with previous run
# but we have to reset the profile attached to the gene gene._profile._report
self.profile_factory = ProfileFactory(self.cfg)
c_gene_abc = CoreGene(self.model_location, gene_name,
self.profile_factory)
report = search_genes([c_gene_abc], self.cfg)
self.assertEqual(len(report), 1)
self.assertEqual(expected_hit[0], report[0].hits[0])
