def test_previous_run(self):
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'previous_run' : 'foo',
'res_search_dir' : self.tmp_dir
}
self.assertRaises(ValueError, Config, **kwargs)
try:
cfg_base = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertIsNone(cfg_base.previous_run)
cfg_base.save( cfg_base.working_dir )
#wait
time.sleep(1)
new_cfg = Config(previous_run = cfg_base.working_dir)
self.assertEqual(new_cfg.previous_run, cfg_base.working_dir)
finally:
# close loggers filehandles, so they don't block file deletion
# in shutil.rmtree calls in Windows
logging.shutdown()
try:
shutil.rmtree(new_cfg.working_dir)
except:
pass
def test_worker_nb(self):
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.worker_nb, 1)
self.tearDown()
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir,
worker_nb = 2
)
self.assertEqual(self.cfg.worker_nb, 2)
self.tearDown()
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'res_search_dir' : self.tmp_dir,
'worker_nb' : '2.3'
}
self.assertRaises(ValueError, Config, **kwargs)
def test_def_dir(self):
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'def_dir': '/kjhdsjkg938268235764kjdsg',
'res_search_dir' : self.tmp_dir
}
real_def_dir = os.path.join(os.path.dirname(__file__), '..', 'data', 'DEF'),
self.assertRaises(ValueError, Config, **kwargs)
def_dir = os.path.join(tempfile.gettempdir(), 'macsyfinder_DEF')
if not os.path.exists(def_dir):
os.mkdir(def_dir)
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = def_dir,
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertEqual(def_dir, self.cfg.def_dir)
shutil.rmtree(def_dir)
self.tearDown()
def_dir = os.path.join(os.path.dirname(__file__),'..', 'data', 'DEF')
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = def_dir,
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertEqual(def_dir, self.cfg.def_dir)
def test_previous_run(self):
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'previous_run' : 'foo',
'res_search_dir' : '/tmp',
}
self.assertRaises(ValueError, Config, **kwargs)
try:
cfg_base = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = '/tmp',
)
self.assertIsNone(cfg_base.previous_run)
cfg_base.save( cfg_base.working_dir )
#wait
time.sleep(1)
new_cfg = Config(previous_run = cfg_base.working_dir)
self.assertEqual(new_cfg.previous_run, cfg_base.working_dir)
finally:
try:
shutil.rmtree(cfg_base.working_dir)
except:
pass
try:
shutil.rmtree(new_cfg.working_dir)
except:
pass
def test_coverage_profile(self):
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir,
)
self.assertEqual( self.cfg.coverage_profile, 0.5 )
self.tearDown()
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
coverage_profile = 0.6,
res_search_dir = self.tmp_dir,
)
self.assertEqual( self.cfg.coverage_profile, 0.6 )
self.tearDown()
#coverage_profile must be a float
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'coverage_profile' : "foo",
'res_search_dir' : self.tmp_dir
}
self.assertRaises(ValueError, Config, **kwargs)
def test_replicon_topology(self):
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.replicon_topology, 'circular')
self.tearDown()
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
replicon_topology = 'linear',
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.replicon_topology, 'linear')
self.tearDown()
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'replicon_topology' : 'foo',
'res_search_dir' : self.tmp_dir
}
self.assertRaises(ValueError, Config, **kwargs)
def test_build_indexes(self):
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertFalse(self.cfg.build_indexes)
self.tearDown()
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'build_indexes' : True
}
config = Config(
cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir,
build_indexes = True
)
self.assertTrue(config.build_indexes)
def test_sequence_db(self):
kwargs = {'cfg_file' : "nimportnaoik",
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'worker_nb' : '2.3' ,
'res_search_dir' : '/tmp',
}
self.assertRaises(ValueError, Config, **kwargs)
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : "foo",
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'worker_nb' : '2.3',
'res_search_dir' : '/tmp',
}
self.assertRaises(ValueError, Config, **kwargs)
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa")
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = sequence_db,
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = '/tmp',
)
self.assertEqual(self.cfg.sequence_db, sequence_db)
def test_sequence_db(self):
kwargs = {'cfg_file' : "nimportnaoik",
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'worker_nb' : '2.3' ,
'res_search_dir' : self.tmp_dir,
'log_file' : 'NUL' if platform.system() == 'Windows' else '/dev/null'
}
self.assertRaises(ValueError, Config, **kwargs)
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : "foo",
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'worker_nb' : '2.3',
'res_search_dir' : self.tmp_dir,
'log_file' : 'NUL' if platform.system() == 'Windows' else '/dev/null'
}
self.assertRaises(ValueError, Config, **kwargs)
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa")
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = sequence_db,
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir,
log_file = 'NUL' if platform.system() == 'Windows' else '/dev/null'
)
self.assertEqual(self.cfg.sequence_db, sequence_db)
def test_default(self):
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.hmmer_exe, 'hmmsearch')
def test_index_db_exe(self):
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.index_db_exe, 'makeblastdb')
self.tearDown()
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
index_db_exe = 'truc',
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.index_db_exe, 'truc')
def test_hmmer_exe(self):
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = '/tmp'
)
self.assertEqual(self.cfg.hmmer_exe, 'hmmsearch')
self.tearDown()
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
hmmer_exe = 'truc',
res_search_dir = '/tmp',
)
self.assertEqual(self.cfg.hmmer_exe, 'truc')
def test_profile_suffix(self):
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.profile_suffix, '.hmm')
self.tearDown()
profile_suffix = 'foo'
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
profile_suffix = profile_suffix,
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.profile_suffix, profile_suffix)
def test_res_extract_suffix(self):
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = '/tmp',
)
self.assertEqual(self.cfg.res_extract_suffix, '.res_hmm_extract')
self.tearDown()
res_extract_suffix = 'foo'
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_extract_suffix = res_extract_suffix,
res_search_dir = '/tmp',
)
self.assertEqual(self.cfg.res_extract_suffix, res_extract_suffix)
def test_res_search_suffix(self):
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(os.path.dirname(__file__),'..', 'data', 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.res_search_suffix, '.search_hmm.out')
self.tearDown()
res_search_suffix = 'foo'
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_suffix = res_search_suffix,
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.res_search_suffix, res_search_suffix)
def test_i_value_sel(self):
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'i_evalue_sel' : 'foo',
'res_search_dir' : self.tmp_dir
}
self.assertRaises(ValueError, Config, **kwargs)
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.i_evalue_sel, 0.001)
self.tearDown()
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
i_evalue_sel = 0.7,
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.i_evalue_sel, 0.7)
self.tearDown()
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'e_value_res' : 0.7,
'i_evalue_sel' : 1,
'res_search_dir' : self.tmp_dir
}
self.assertRaises(ValueError, Config, **kwargs)
def test_min_mandatory_genes_required(self):
min_mandatory_genes_required = (["T2SS", 32], ['Flagellum', 64])
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
min_mandatory_genes_required = min_mandatory_genes_required,
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.min_mandatory_genes_required('T2SS'), 32)
self.assertEqual(self.cfg.min_mandatory_genes_required('Flagellum'), 64)
self.assertIsNone(self.cfg.min_mandatory_genes_required('Foo'))
def test_multi_loci(self):
multi_loci = "T2SS,Flagellum"
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
multi_loci = multi_loci,
res_search_dir = self.tmp_dir
)
self.assertTrue(self.cfg.multi_loci('T2SS'))
self.assertTrue(self.cfg.multi_loci('Flagellum'))
self.assertFalse(self.cfg.multi_loci('Foo'))
def test_max_nb_genes(self):
max_nb_genes = (["T2SS", 32], ['Flagellum', 64])
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
max_nb_genes = max_nb_genes,
res_search_dir = '/tmp',
)
self.assertEqual(self.cfg.max_nb_genes('T2SS'), 32)
self.assertEqual(self.cfg.max_nb_genes('Flagellum'), 64)
self.assertIsNone(self.cfg.max_nb_genes('Foo'))
def test_inter_gene_max_space(self):
inter_gene_max_space = (["T2SS", 32], ['Flagellum', 64])
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
inter_gene_max_space = inter_gene_max_space,
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.inter_gene_max_space('T2SS'), 32)
self.assertEqual(self.cfg.inter_gene_max_space('Flagellum'), 64)
self.assertIsNone(self.cfg.inter_gene_max_space('Foo'))
def test_profile_dir(self):
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : 'foo',
'res_search_dir' : self.tmp_dir
}
self.assertRaises(ValueError, Config, **kwargs)
profile_dir = os.path.join(os.path.dirname(__file__), '..', 'data', 'profiles')
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = profile_dir,
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.profile_dir, profile_dir)
def test_db_type(self):
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = '/tmp',
)
self.assertEqual( self.cfg.db_type, 'gembase')
self.tearDown()
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'foo',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'res_search_dir' : '/tmp',
}
self.assertRaises(ValueError, Config, **kwargs)
def test_res_search_dir(self):
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'res_search_dir' : 'foo',
'log_file' : 'NUL' if platform.system() == 'Windows' else '/dev/null'
}
self.assertRaises(ValueError, Config, **kwargs)
self.tearDown()
res_search_dir = os.path.join(os.path.dirname(__file__), 'datatest')
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = res_search_dir,
log_file = 'NUL' if platform.system() == 'Windows' else '/dev/null'
)
self.assertEqual(self.cfg.res_search_dir, res_search_dir)
def test_models_dir(self):
self.parsed_args.models_dir = self.tmp_dir
cfg = Config(self.defaults, self.parsed_args)
self.assertEqual(cfg.models_dir(), [self.tmp_dir])
def test_working_dir(self):
cfg = Config(self.defaults, self.parsed_args)
self.assertEqual(cfg.out_dir(), cfg.working_dir())
def test_bad_topology(self):
self.parsed_args.replicon_topology = "FOO"
with self.assertRaises(ValueError) as ctx:
Config(self.defaults, self.parsed_args)
self.assertEqual(str(ctx.exception),
"replicon_topology as unauthorized value : 'FOO'.")
def test_bad_db_type(self):
self.parsed_args.db_type = "FOO"
with self.assertRaises(ValueError) as ctx:
Config(self.defaults, self.parsed_args)
self.assertEqual(str(ctx.exception),
"db_type as unauthorized value : 'FOO'.")
def setUp(self):
self.args = argparse.Namespace()
self.args.db_type = 'gembase'
self.args.models_dir = self.find_data('models')
self.args.res_search_dir = tempfile.gettempdir()
self.args.log_level = 30
self.args.out_dir = os.path.join(self.args.res_search_dir,
'test_macsyfinder_repliconDB')
if os.path.exists(self.args.out_dir):
shutil.rmtree(self.args.out_dir)
os.mkdir(self.args.out_dir)
seq_db = self.find_data("base", "test_base.fa")
shutil.copy(seq_db, self.args.out_dir)
self.args.sequence_db = os.path.join(self.args.out_dir,
os.path.basename(seq_db))
self.cfg = Config(MacsyDefaults(), self.args)
self.ESCO030p01_genes = [('000010', '886'), ('000020', '291'),
('000030', '656'), ('000040', '500'),
('000050', '407'), ('000060', '144'),
('000070', '183'), ('000080', '121'),
('000090', '199'), ('000100', '325'),
('000110', '425'), ('000120', '171'),
('000130', '277'), ('000140', '133'),
('000150', '108'), ('000160', '295'),
('000170', '273'), ('000180', '367'),
('000190', '573'), ('000200', '343'),
('000210', '295'), ('000220', '108'),
('000230', '117'), ('000240', '153'),
('000250', '479'), ('000260', '706'),
('000270', '998'), ('000280', '171'),
('000290', '108'), ('000300', '295'),
('000310', '165'), ('000320', '243'),
('000330', '295'), ('000340', '108'),
('000350', '1755'), ('000360', '248'),
('000370', '286'), ('000380', '186'),
('000390', '83'), ('000400', '153'),
('000410', '69'), ('000420', '295'),
('000430', '108'), ('000440', '145'),
('000450', '59'), ('000460', '124'),
('000470', '246'), ('000480', '325'),
('000490', '54'), ('000500', '95'),
('000510', '83'), ('000520', '56'),
('000530', '401'), ('000540', '320'),
('000550', '256'), ('000560', '73'),
('000570', '144'), ('000580', '258'),
('000590', '133'), ('000600', '140'),
('000610', '63'), ('000620', '138'),
('000630', '68'), ('000640', '169'),
('000650', '127'), ('000660', '295'),
('000670', '108'), ('000670', '108')]
self.PSAE001c01_genes = [('006940', '803'), ('013980', '759'),
('017350', '600'), ('018920', '776'),
('026600', '273'), ('031420', '658'),
('043580', '416'), ('051090', '714'),
('055870', '449'), ('055880', '447'),
('055890', '588'), ('055900', '292'),
('055910', '262'), ('055920', '166'),
('055930', '288'), ('055940', '194'),
('055950', '567'), ('055960', '188'),
('055970', '247'), ('055980', '252'),
('055990', '455'), ('056000', '450'),
('056010', '260'), ('056020', '246'),
('056030', '70'), ('056040', '133'),
('056050', '284'), ('056060', '585'),
('056070', '435'), ('056080', '342'),
('056090', '252'), ('056100', '122'),
('056110', '213'), ('056120', '400'),
('056130', '134'), ('056140', '138'),
('056150', '397'), ('056160', '298'),
('056170', '186'), ('056180', '445'),
('056190', '414'), ('056200', '132'),
('056210', '674'), ('056220', '319'),
('056230', '394'), ('056240', '207'),
('056250', '401'), ('056260', '611'),
('056270', '257'), ('056280', '169'),
('056290', '454'), ('056300', '141'),
('056310', '458'), ('056320', '286'),
('056330', '514'), ('056340', '178'),
('056350', '156'), ('056360', '85'),
('056370', '289'), ('056380', '126'),
('056390', '290'), ('056400', '262'),
('056410', '214'), ('056420', '630'),
('056430', '127'), ('056440', '455'),
('056440', '455')]
self.NCDB_genes = [('056134', '289'), ('056135', '126'),
('056136', '290'), ('056137', '262'),
('056138', '214'), ('056139', '630'),
('056140', '127'), ('056141', '803'),
('056141', '803')]
idx = Indexes(self.cfg)
idx._build_my_indexes()
class Test(MacsyTest):
def __init__(self, methodName='runTest'):
super(Test, self).__init__(methodName)
def fake_init(obj, cfg):
obj.cfg = cfg
idx = Indexes(cfg)
obj.sequence_idx = idx.find_my_indexes()
obj.topology_file = cfg.topology_file()
obj._DB = {}
self.fake_init = fake_init
self.real_init = RepliconDB.__init__
def setUp(self):
self.args = argparse.Namespace()
self.args.db_type = 'gembase'
self.args.models_dir = self.find_data('models')
self.args.res_search_dir = tempfile.gettempdir()
self.args.log_level = 30
self.args.out_dir = os.path.join(self.args.res_search_dir,
'test_macsyfinder_repliconDB')
if os.path.exists(self.args.out_dir):
shutil.rmtree(self.args.out_dir)
os.mkdir(self.args.out_dir)
seq_db = self.find_data("base", "test_base.fa")
shutil.copy(seq_db, self.args.out_dir)
self.args.sequence_db = os.path.join(self.args.out_dir,
os.path.basename(seq_db))
self.cfg = Config(MacsyDefaults(), self.args)
self.ESCO030p01_genes = [('000010', '886'), ('000020', '291'),
('000030', '656'), ('000040', '500'),
('000050', '407'), ('000060', '144'),
('000070', '183'), ('000080', '121'),
('000090', '199'), ('000100', '325'),
('000110', '425'), ('000120', '171'),
('000130', '277'), ('000140', '133'),
('000150', '108'), ('000160', '295'),
('000170', '273'), ('000180', '367'),
('000190', '573'), ('000200', '343'),
('000210', '295'), ('000220', '108'),
('000230', '117'), ('000240', '153'),
('000250', '479'), ('000260', '706'),
('000270', '998'), ('000280', '171'),
('000290', '108'), ('000300', '295'),
('000310', '165'), ('000320', '243'),
('000330', '295'), ('000340', '108'),
('000350', '1755'), ('000360', '248'),
('000370', '286'), ('000380', '186'),
('000390', '83'), ('000400', '153'),
('000410', '69'), ('000420', '295'),
('000430', '108'), ('000440', '145'),
('000450', '59'), ('000460', '124'),
('000470', '246'), ('000480', '325'),
('000490', '54'), ('000500', '95'),
('000510', '83'), ('000520', '56'),
('000530', '401'), ('000540', '320'),
('000550', '256'), ('000560', '73'),
('000570', '144'), ('000580', '258'),
('000590', '133'), ('000600', '140'),
('000610', '63'), ('000620', '138'),
('000630', '68'), ('000640', '169'),
('000650', '127'), ('000660', '295'),
('000670', '108'), ('000670', '108')]
self.PSAE001c01_genes = [('006940', '803'), ('013980', '759'),
('017350', '600'), ('018920', '776'),
('026600', '273'), ('031420', '658'),
('043580', '416'), ('051090', '714'),
('055870', '449'), ('055880', '447'),
('055890', '588'), ('055900', '292'),
('055910', '262'), ('055920', '166'),
('055930', '288'), ('055940', '194'),
('055950', '567'), ('055960', '188'),
('055970', '247'), ('055980', '252'),
('055990', '455'), ('056000', '450'),
('056010', '260'), ('056020', '246'),
('056030', '70'), ('056040', '133'),
('056050', '284'), ('056060', '585'),
('056070', '435'), ('056080', '342'),
('056090', '252'), ('056100', '122'),
('056110', '213'), ('056120', '400'),
('056130', '134'), ('056140', '138'),
('056150', '397'), ('056160', '298'),
('056170', '186'), ('056180', '445'),
('056190', '414'), ('056200', '132'),
('056210', '674'), ('056220', '319'),
('056230', '394'), ('056240', '207'),
('056250', '401'), ('056260', '611'),
('056270', '257'), ('056280', '169'),
('056290', '454'), ('056300', '141'),
('056310', '458'), ('056320', '286'),
('056330', '514'), ('056340', '178'),
('056350', '156'), ('056360', '85'),
('056370', '289'), ('056380', '126'),
('056390', '290'), ('056400', '262'),
('056410', '214'), ('056420', '630'),
('056430', '127'), ('056440', '455'),
('056440', '455')]
self.NCDB_genes = [('056134', '289'), ('056135', '126'),
('056136', '290'), ('056137', '262'),
('056138', '214'), ('056139', '630'),
('056140', '127'), ('056141', '803'),
('056141', '803')]
idx = Indexes(self.cfg)
idx._build_my_indexes()
def tearDown(self):
try:
shutil.rmtree(self.cfg.working_dir())
except:
pass
RepliconDB.__init__ = self.real_init
def test_fill_topology(self):
self.args.topology_file = self.args.sequence_db + ".topo"
db_send = {'ESCO030p01': 'circular', 'PSAE001c01': 'linear'}
with open(self.args.topology_file, 'w') as f:
for k, v in list(db_send.items()):
f.write('{0} : {1}\n'.format(k, v))
cfg = Config(MacsyDefaults(), self.args)
RepliconDB.__init__ = self.fake_init
db = RepliconDB(cfg)
rcv_topo = db._fill_topology()
self.assertDictEqual(db_send, rcv_topo)
def test_fill_ordered_replicon_min_max(self):
seq_ori = self.find_data("base", "ordered_replicon_base.fasta")
shutil.copy(seq_ori, self.args.out_dir)
self.args.sequence_db = os.path.join(self.args.out_dir,
os.path.basename(seq_ori))
cfg = Config(MacsyDefaults(), self.args)
idx = Indexes(cfg)
idx._build_my_indexes()
RepliconDB.__init__ = self.fake_init
db = RepliconDB(cfg)
db._fill_ordered_min_max(cfg.replicon_topology())
self.assertEqual(len(db._DB), 1)
rep = db[RepliconDB.ordered_replicon_name]
self.assertEqual(rep.topology, cfg.replicon_topology())
self.assertEqual(rep.min, 1)
self.assertEqual(rep.max, 52)
def test_fill_gembase_min_max_default_topology(self):
RepliconDB.__init__ = self.fake_init
db = RepliconDB(self.cfg)
db._fill_gembase_min_max({}, self.cfg.replicon_topology())
self.assertEqual(len(db._DB), 3)
self.assertEqual(set(db._DB.keys()),
set(['ESCO030p01', 'PSAE001c01', 'NC_xxxxx_xx']))
PRRU001c01 = db['ESCO030p01']
self.assertEqual(PRRU001c01.topology, 'circular')
self.assertEqual(PRRU001c01.min, 1)
self.assertEqual(PRRU001c01.max, 67)
self.assertEqual(PRRU001c01.genes, self.ESCO030p01_genes)
PSAE001c01 = db['PSAE001c01']
self.assertEqual(PSAE001c01.topology, 'circular')
self.assertEqual(PSAE001c01.min, 68)
self.assertEqual(PSAE001c01.max, 133)
self.assertEqual(PSAE001c01.genes, self.PSAE001c01_genes)
DBNC = db['NC_xxxxx_xx']
self.assertEqual(DBNC.topology, 'circular')
self.assertEqual(DBNC.min, 134)
self.assertEqual(DBNC.max, 141)
self.assertEqual(DBNC.genes, self.NCDB_genes)
def test_fill_gembase_min_max_with_topology(self):
self.args.topology_file = self.args.sequence_db + ".topo"
with open(self.args.topology_file, 'w') as f:
f.write(
'# topology file\nESCO030p01 : circular\nPSAE001c01 : linear\n'
)
cfg = Config(MacsyDefaults(), self.args)
RepliconDB.__init__ = self.fake_init
db = RepliconDB(cfg)
topo_dict = db._fill_topology()
db._fill_gembase_min_max(topo_dict, 'circular')
self.assertEqual(len(db._DB), 3)
self.assertEqual(set(db._DB.keys()),
set(['ESCO030p01', 'PSAE001c01', 'NC_xxxxx_xx']))
ESCO030p01 = db['ESCO030p01']
self.assertEqual(ESCO030p01.topology, 'circular')
self.assertEqual(ESCO030p01.min, 1)
self.assertEqual(ESCO030p01.max, 67)
self.assertEqual(ESCO030p01.genes, self.ESCO030p01_genes)
PSAE001c01 = db['PSAE001c01']
self.assertEqual(PSAE001c01.topology, 'linear')
self.assertEqual(PSAE001c01.min, 68)
self.assertEqual(PSAE001c01.max, 133)
self.assertEqual(PSAE001c01.genes, self.PSAE001c01_genes)
DBNC = db['NC_xxxxx_xx']
self.assertEqual(DBNC.topology, 'circular')
self.assertEqual(DBNC.min, 134)
self.assertEqual(DBNC.max, 141)
self.assertEqual(DBNC.genes, self.NCDB_genes)
def test_in(self):
db = RepliconDB(self.cfg)
self.assertIn('ESCO030p01', db)
self.assertIn('PSAE001c01', db)
self.assertIn('NC_xxxxx_xx', db)
self.assertNotIn('toto', db)
def test_getitem(self):
db = RepliconDB(self.cfg)
ESCO030p01 = RepliconInfo(self.cfg.replicon_topology(), 1, 67,
self.ESCO030p01_genes)
PSAE001c01 = RepliconInfo(self.cfg.replicon_topology(), 68, 133,
self.PSAE001c01_genes)
NCXX = RepliconInfo("circular", 134, 141, self.NCDB_genes)
self.assertEqual(ESCO030p01, db['ESCO030p01'])
self.assertEqual(PSAE001c01, db['PSAE001c01'])
self.assertEqual(NCXX, db['NC_xxxxx_xx'])
self.assertRaises(KeyError, db.__getitem__, 'foo')
def test_get(self):
db = RepliconDB(self.cfg)
ESCO030p01 = RepliconInfo(self.cfg.replicon_topology(), 1, 67,
self.ESCO030p01_genes)
PSAE001c01 = RepliconInfo(self.cfg.replicon_topology(), 68, 133,
self.PSAE001c01_genes)
NCXX = RepliconInfo("circular", 134, 141, self.NCDB_genes)
self.assertEqual(ESCO030p01, db.get('ESCO030p01'))
self.assertEqual(PSAE001c01, db.get('PSAE001c01'))
self.assertEqual(NCXX, db.get('NC_xxxxx_xx', 'foo'))
self.assertIsNone(db.get('foo'))
self.assertEqual('bar', db.get('foo', 'bar'))
def test_items(self):
db = RepliconDB(self.cfg)
ESCO030p01 = RepliconInfo(self.cfg.replicon_topology(), 1, 67,
self.ESCO030p01_genes)
PSAE001c01 = RepliconInfo(self.cfg.replicon_topology(), 68, 133,
self.PSAE001c01_genes)
NCXX = RepliconInfo("circular", 134, 141, self.NCDB_genes)
self.assertCountEqual(list(db.items()), [('ESCO030p01', ESCO030p01),
('NC_xxxxx_xx', NCXX),
('PSAE001c01', PSAE001c01)])
def test_iteritems(self):
db = RepliconDB(self.cfg)
ESCO030p01 = RepliconInfo(self.cfg.replicon_topology(), 1, 67,
self.ESCO030p01_genes)
PSAE001c01 = RepliconInfo(self.cfg.replicon_topology(), 68, 133,
self.PSAE001c01_genes)
NCXX = RepliconInfo("circular", 134, 141, self.NCDB_genes)
self.assertCountEqual(iter(db.items()), [('ESCO030p01', ESCO030p01),
('NC_xxxxx_xx', NCXX),
('PSAE001c01', PSAE001c01)])
def test_replicon_infos(self):
db = RepliconDB(self.cfg)
ESCO030p01 = RepliconInfo(self.cfg.replicon_topology(), 1, 67,
self.ESCO030p01_genes)
PSAE001c01 = RepliconInfo(self.cfg.replicon_topology(), 68, 133,
self.PSAE001c01_genes)
NCXX = RepliconInfo("circular", 134, 141, self.NCDB_genes)
values = db.replicon_infos()
self.assertCountEqual(values, [ESCO030p01, NCXX, PSAE001c01])
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])
class Test(unittest.TestCase):
_data_dir = os.path.join(os.path.dirname(__file__), "datatest")
def setUp(self):
l = logging.getLogger()
l.manager.loggerDict.clear()
if hasattr(self, 'tmp_dir'):
try:
shutil.rmtree(self.cfg.working_dir)
except Exception as err:
pass
self.tmp_dir = tempfile.gettempdir()
def tearDown(self):
logging.shutdown()
l = logging.getLogger()
l.manager.loggerDict.clear()
if hasattr(self, 'cfg'):
try:
shutil.rmtree(self.cfg.working_dir)
except Exception as err:
pass
def test_build_indexes(self):
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertFalse(self.cfg.build_indexes)
self.tearDown()
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'build_indexes' : True
}
config = Config(
cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir,
build_indexes = True
)
self.assertTrue(config.build_indexes)
def test_default(self):
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.hmmer_exe, 'hmmsearch')
def test_coverage_profile(self):
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir,
)
self.assertEqual( self.cfg.coverage_profile, 0.5 )
self.tearDown()
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
coverage_profile = 0.6,
res_search_dir = self.tmp_dir,
)
self.assertEqual( self.cfg.coverage_profile, 0.6 )
self.tearDown()
#coverage_profile must be a float
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'coverage_profile' : "foo",
'res_search_dir' : self.tmp_dir
}
self.assertRaises(ValueError, Config, **kwargs)
def test_def_dir(self):
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'def_dir': '/kjhdsjkg938268235764kjdsg',
'res_search_dir' : self.tmp_dir
}
real_def_dir = os.path.join(os.path.dirname(__file__), '..', 'data', 'DEF'),
self.assertRaises(ValueError, Config, **kwargs)
def_dir = os.path.join(tempfile.gettempdir(), 'macsyfinder_DEF')
if not os.path.exists(def_dir):
os.mkdir(def_dir)
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = def_dir,
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertEqual(def_dir, self.cfg.def_dir)
shutil.rmtree(def_dir)
self.tearDown()
def_dir = os.path.join(os.path.dirname(__file__),'..', 'data', 'DEF')
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = def_dir,
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertEqual(def_dir, self.cfg.def_dir)
def test_e_value_res(self):
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'e_value_res' : 'foo',
'res_search_dir' : self.tmp_dir
}
self.assertRaises(ValueError, Config, **kwargs)
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.e_value_res, 1)
self.tearDown()
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
e_value_res = 0.7,
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.e_value_res, 0.7)
self.tearDown()
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'e_value_res' : 0.7,
'i_evalue_sel' : 1,
'res_search_dir' : self.tmp_dir
}
self.assertRaises(ValueError, Config, **kwargs)
def test_hmmer_exe(self):
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.hmmer_exe, 'hmmsearch')
self.tearDown()
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
hmmer_exe = 'truc',
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.hmmer_exe, 'truc')
def test_index_db_exe(self):
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.index_db_exe, 'makeblastdb')
self.tearDown()
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
index_db_exe = 'truc',
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.index_db_exe, 'truc')
def test_i_value_sel(self):
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'i_evalue_sel' : 'foo',
'res_search_dir' : self.tmp_dir
}
self.assertRaises(ValueError, Config, **kwargs)
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.i_evalue_sel, 0.001)
self.tearDown()
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
i_evalue_sel = 0.7,
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.i_evalue_sel, 0.7)
self.tearDown()
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'e_value_res' : 0.7,
'i_evalue_sel' : 1,
'res_search_dir' : self.tmp_dir
}
self.assertRaises(ValueError, Config, **kwargs)
def test_db_type(self):
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertEqual( self.cfg.db_type, 'gembase')
self.tearDown()
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'foo',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'res_search_dir' : self.tmp_dir
}
self.assertRaises(ValueError, Config, **kwargs)
def test_previous_run(self):
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'previous_run' : 'foo',
'res_search_dir' : self.tmp_dir
}
self.assertRaises(ValueError, Config, **kwargs)
try:
cfg_base = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertIsNone(cfg_base.previous_run)
cfg_base.save( cfg_base.working_dir )
#wait
time.sleep(1)
new_cfg = Config(previous_run = cfg_base.working_dir)
self.assertEqual(new_cfg.previous_run, cfg_base.working_dir)
finally:
# close loggers filehandles, so they don't block file deletion
# in shutil.rmtree calls in Windows
logging.shutdown()
try:
shutil.rmtree(new_cfg.working_dir)
except:
pass
def test_profile_dir(self):
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : 'foo',
'res_search_dir' : self.tmp_dir
}
self.assertRaises(ValueError, Config, **kwargs)
profile_dir = os.path.join(os.path.dirname(__file__), '..', 'data', 'profiles')
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = profile_dir,
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.profile_dir, profile_dir)
def test_profile_suffix(self):
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.profile_suffix, '.hmm')
self.tearDown()
profile_suffix = 'foo'
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
profile_suffix = profile_suffix,
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.profile_suffix, profile_suffix)
def test_replicon_topology(self):
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.replicon_topology, 'circular')
self.tearDown()
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
replicon_topology = 'linear',
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.replicon_topology, 'linear')
self.tearDown()
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'replicon_topology' : 'foo',
'res_search_dir' : self.tmp_dir
}
self.assertRaises(ValueError, Config, **kwargs)
def test_inter_gene_max_space(self):
inter_gene_max_space = (["T2SS", 32], ['Flagellum', 64])
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
inter_gene_max_space = inter_gene_max_space,
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.inter_gene_max_space('T2SS'), 32)
self.assertEqual(self.cfg.inter_gene_max_space('Flagellum'), 64)
self.assertIsNone(self.cfg.inter_gene_max_space('Foo'))
def test_min_genes_required(self):
min_genes_required = (["T2SS", 32], ['Flagellum', 64])
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
min_genes_required = min_genes_required,
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.min_genes_required('T2SS'), 32)
self.assertEqual(self.cfg.min_genes_required('Flagellum'), 64)
self.assertIsNone(self.cfg.min_genes_required('Foo'))
def test_max_nb_genes(self):
max_nb_genes = (["T2SS", 32], ['Flagellum', 64])
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
max_nb_genes = max_nb_genes,
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.max_nb_genes('T2SS'), 32)
self.assertEqual(self.cfg.max_nb_genes('Flagellum'), 64)
self.assertIsNone(self.cfg.max_nb_genes('Foo'))
def test_min_mandatory_genes_required(self):
min_mandatory_genes_required = (["T2SS", 32], ['Flagellum', 64])
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
min_mandatory_genes_required = min_mandatory_genes_required,
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.min_mandatory_genes_required('T2SS'), 32)
self.assertEqual(self.cfg.min_mandatory_genes_required('Flagellum'), 64)
self.assertIsNone(self.cfg.min_mandatory_genes_required('Foo'))
def test_multi_loci(self):
multi_loci = "T2SS,Flagellum"
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
multi_loci = multi_loci,
res_search_dir = self.tmp_dir
)
self.assertTrue(self.cfg.multi_loci('T2SS'))
self.assertTrue(self.cfg.multi_loci('Flagellum'))
self.assertFalse(self.cfg.multi_loci('Foo'))
def test_res_extract_suffix(self):
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.res_extract_suffix, '.res_hmm_extract')
self.tearDown()
res_extract_suffix = 'foo'
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_extract_suffix = res_extract_suffix,
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.res_extract_suffix, res_extract_suffix)
def test_res_search_dir(self):
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'res_search_dir' : 'foo',
'log_file' : 'NUL' if platform.system() == 'Windows' else '/dev/null'
}
self.assertRaises(ValueError, Config, **kwargs)
self.tearDown()
res_search_dir = os.path.join(os.path.dirname(__file__), 'datatest')
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = res_search_dir,
log_file = 'NUL' if platform.system() == 'Windows' else '/dev/null'
)
self.assertEqual(self.cfg.res_search_dir, res_search_dir)
def test_res_search_suffix(self):
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(os.path.dirname(__file__),'..', 'data', 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.res_search_suffix, '.search_hmm.out')
self.tearDown()
res_search_suffix = 'foo'
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_suffix = res_search_suffix,
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.res_search_suffix, res_search_suffix)
def test_sequence_db(self):
kwargs = {'cfg_file' : "nimportnaoik",
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'worker_nb' : '2.3' ,
'res_search_dir' : self.tmp_dir,
'log_file' : 'NUL' if platform.system() == 'Windows' else '/dev/null'
}
self.assertRaises(ValueError, Config, **kwargs)
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : "foo",
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'worker_nb' : '2.3',
'res_search_dir' : self.tmp_dir,
'log_file' : 'NUL' if platform.system() == 'Windows' else '/dev/null'
}
self.assertRaises(ValueError, Config, **kwargs)
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa")
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = sequence_db,
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir,
log_file = 'NUL' if platform.system() == 'Windows' else '/dev/null'
)
self.assertEqual(self.cfg.sequence_db, sequence_db)
def test_worker_nb(self):
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir
)
self.assertEqual(self.cfg.worker_nb, 1)
self.tearDown()
self.cfg = Config(cfg_file = "nimportnaoik",
sequence_db = os.path.join(self._data_dir, "base", "test_base.fa"),
db_type = 'gembase',
def_dir = os.path.join(self._data_dir, 'DEF'),
profile_dir = os.path.join(self._data_dir, 'profiles'),
res_search_dir = self.tmp_dir,
worker_nb = 2
)
self.assertEqual(self.cfg.worker_nb, 2)
self.tearDown()
kwargs = {'cfg_file' : "nimportnaoik",
'sequence_db' : os.path.join(self._data_dir, "base", "test_base.fa"),
'db_type' : 'gembase',
'def_dir' : os.path.join(self._data_dir, 'DEF'),
'profile_dir' : os.path.join(self._data_dir, 'profiles'),
'res_search_dir' : self.tmp_dir,
'worker_nb' : '2.3'
}
self.assertRaises(ValueError, Config, **kwargs)
