def test_fill_ordered_replicon_min_max(self):
self.tearDown()
self.cfg = Config(hmmer_exe="hmmsearch",
sequence_db=os.path.join(self._data_dir, "base",
"ordered_replicon_base"),
db_type="ordered_replicon",
e_value_res=1,
i_evalue_sel=0.5,
def_dir=os.path.join(self._data_dir, 'DEF'),
res_search_dir='/tmp',
res_search_suffix=".search_hmm.out",
profile_dir=os.path.join(self._data_dir, 'profiles'),
profile_suffix=".hmm",
res_extract_suffix="",
log_level=30,
log_file='/dev/null')
shutil.copy(self.cfg.sequence_db, self.cfg.working_dir)
self.cfg.options['sequence_db'] = os.path.join(
self.cfg.working_dir, os.path.basename(self.cfg.sequence_db))
idx = Indexes(self.cfg)
idx._build_my_indexes()
RepliconDB.__init__ = self.fake_init
db = RepliconDB(self.cfg)
db._fill_ordered_min_max(self.cfg.replicon_topology)
self.assertEqual(len(db._DB), 1)
rep = db[RepliconDB.ordered_replicon_name]
self.assertEqual(rep.topology, self.cfg.replicon_topology)
self.assertEqual(rep.min, 1)
self.assertEqual(rep.max, 52)
def test_fill_gembase_min_max_with_topology(self):
self.cfg.options['topology_file'] = self.cfg.sequence_db + ".topo"
with open(self.cfg.topology_file , 'w') as f:
f.write('# topology file\nESCO030p01 : circular\nPSAE001c01 : linear\n')
RepliconDB.__init__ = self.fake_init
db = RepliconDB(self.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_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.assertItemsEqual(db.items(), [('ESCO030p01',ESCO030p01), ('NC_xxxxx_xx',NCXX),
('PSAE001c01',PSAE001c01)])
def test_fill_ordered_replicon_min_max(self):
self.tearDown()
self.cfg = Config( hmmer_exe = "hmmsearch",
sequence_db = os.path.join(self._data_dir, "base", "ordered_replicon_base"),
db_type = "ordered_replicon",
e_value_res = 1,
i_evalue_sel = 0.5,
def_dir = os.path.join(self._data_dir, 'DEF'),
res_search_dir = tempfile.gettempdir(),
res_search_suffix = ".search_hmm.out",
profile_dir = os.path.join(self._data_dir, 'profiles'),
profile_suffix = ".hmm",
res_extract_suffix = "",
log_level = 30,
log_file = 'NUL' if platform.system() == 'Windows' else '/dev/null'
)
shutil.copy(self.cfg.sequence_db, self.cfg.working_dir)
self.cfg.options['sequence_db'] = os.path.join(self.cfg.working_dir, os.path.basename(self.cfg.sequence_db))
idx = Indexes(self.cfg)
idx._build_my_indexes()
RepliconDB.__init__ = self.fake_init
db = RepliconDB(self.cfg)
db._fill_ordered_min_max(self.cfg.replicon_topology)
self.assertEqual(len(db._DB), 1)
rep = db[RepliconDB.ordered_replicon_name]
self.assertEqual(rep.topology, self.cfg.replicon_topology)
self.assertEqual(rep.min, 1)
self.assertEqual(rep.max, 52)
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)
self.assertEqual(ESCO030p01, db.get('ESCO030p01'))
self.assertEqual(PSAE001c01, db.get('PSAE001c01'))
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)
self.assertItemsEqual(db.items(), [('ESCO030p01',ESCO030p01),('PSAE001c01',PSAE001c01)])
db = RepliconDB(self.cfg)
PRRU001c01 = RepliconInfo(self.cfg.replicon_topology, 1, 67, self.ESCO030p01_genes)
PSAE001c01 = RepliconInfo(self.cfg.replicon_topology, 68, 133, self.PSAE001c01_genes)
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])
def test_fill_topology(self):
self.cfg.options['topology_file'] = self.cfg.sequence_db + ".topo"
db_send = {'ESCO030p01': 'circular', 'PSAE001c01': 'linear'}
with open(self.cfg.topology_file, 'w') as f:
for k, v in db_send.items():
f.write('%s : %s\n' % (k, v))
RepliconDB.__init__ = self.fake_init
db = RepliconDB(self.cfg)
rcv_topo = db._fill_topology()
self.assertDictEqual(db_send, rcv_topo)
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)
self.assertEqual(ESCO030p01, db.get('ESCO030p01'))
self.assertEqual(PSAE001c01, db.get('PSAE001c01'))
self.assertIsNone(db.get('foo'))
self.assertEqual('bar', db.get('foo', 'bar'))
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)
iter_items = db.iteritems()
for item in [('ESCO030p01', ESCO030p01), ('PSAE001c01', PSAE001c01),
('NC_xxxxx_xx', NCXX)]:
with self.subTest(item=item):
self.assertEqual(next(iter_items), item)
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_topology(self):
self.cfg.options['topology_file'] = self.cfg.sequence_db + ".topo"
db_send = {'ESCO030p01' : 'circular',
'PSAE001c01' : 'linear'
}
with open(self.cfg.topology_file , 'w') as f:
for k, v in db_send.items():
f.write('%s : %s\n' % (k,v))
RepliconDB.__init__ = self.fake_init
db = RepliconDB(self.cfg)
rcv_topo = db._fill_topology()
self.assertDictEqual(db_send, rcv_topo)
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)
self.assertItemsEqual(db.items(), [('ESCO030p01', ESCO030p01),
('PSAE001c01', PSAE001c01)])
db = RepliconDB(self.cfg)
PRRU001c01 = RepliconInfo(self.cfg.replicon_topology, 1, 67,
self.ESCO030p01_genes)
PSAE001c01 = RepliconInfo(self.cfg.replicon_topology, 68, 133,
self.PSAE001c01_genes)
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), 2)
PRRU001c01 = db['ESCO030p01']
self.assertEqual(PRRU001c01.topology, 'circular')
self.assertEqual(PRRU001c01.min, 1)
self.assertEqual(PRRU001c01.max, 67)
PSAE001c01 = db['PSAE001c01']
self.assertEqual(PSAE001c01.topology, 'circular')
self.assertEqual(PSAE001c01.min, 68)
self.assertEqual(PSAE001c01.max, 133)
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), 2)
PRRU001c01 = db['ESCO030p01']
self.assertEqual(PRRU001c01.topology, 'circular')
self.assertEqual(PRRU001c01.min, 1)
self.assertEqual(PRRU001c01.max, 67)
PSAE001c01 = db['PSAE001c01']
self.assertEqual(PSAE001c01.topology, 'circular')
self.assertEqual(PSAE001c01.min, 68)
self.assertEqual(PSAE001c01.max, 133)
def test_fill_gembase_min_max_with_topology(self):
self.cfg.options['topology_file'] = self.cfg.sequence_db + ".topo"
with open(self.cfg.topology_file, 'w') as f:
f.write('ESCO030p01 : circular\nPSAE001c01 : linear\n')
RepliconDB.__init__ = self.fake_init
db = RepliconDB(self.cfg)
topo_dict = db._fill_topology()
db._fill_gembase_min_max(topo_dict, 'circular')
self.assertEqual(len(db._DB), 2)
ESCO030p01 = db['ESCO030p01']
self.assertEqual(ESCO030p01.topology, 'circular')
self.assertEqual(ESCO030p01.min, 1)
self.assertEqual(ESCO030p01.max, 67)
PSAE001c01 = db['PSAE001c01']
self.assertEqual(PSAE001c01.topology, 'linear')
self.assertEqual(PSAE001c01.min, 68)
self.assertEqual(PSAE001c01.max, 133)
def test_fill_gembase_min_max_with_topology(self):
self.cfg.options['topology_file'] = self.cfg.sequence_db + ".topo"
with open(self.cfg.topology_file , 'w') as f:
f.write('ESCO030p01 : circular\nPSAE001c01 : linear\n')
RepliconDB.__init__ = self.fake_init
db = RepliconDB(self.cfg)
topo_dict = db._fill_topology()
db._fill_gembase_min_max(topo_dict, 'circular')
self.assertEqual(len(db._DB), 2)
ESCO030p01 = db['ESCO030p01']
self.assertEqual(ESCO030p01.topology, 'circular')
self.assertEqual(ESCO030p01.min, 1)
self.assertEqual(ESCO030p01.max, 67)
PSAE001c01 = db['PSAE001c01']
self.assertEqual(PSAE001c01.topology, 'linear')
self.assertEqual(PSAE001c01.min, 68)
self.assertEqual(PSAE001c01.max, 133)
def _search_in_ordered_replicon(hits_by_replicon, models_to_detect, config,
logger):
systems = []
rejected_clusters = []
rep_db = RepliconDB(config)
for rep_name in hits_by_replicon:
logger.info(
"\n{:#^60}".format(f" Hits analysis for replicon {rep_name} "))
rep_info = rep_db[rep_name]
for model in models_to_detect:
logger.info(f"Check model {model.fqn}")
hits_related_one_model = model.filter(hits_by_replicon[rep_name])
logger.debug("{:#^80}".format(" hits related to {} ".format(
model.name)))
logger.debug("".join([str(h) for h in hits_related_one_model]))
logger.debug("#" * 80)
logger.info("Building clusters")
hit_weights = HitWeight(**config.hit_weights())
clusters = cluster.build_clusters(hits_related_one_model, rep_info,
model, hit_weights)
logger.debug("{:#^80}".format("CLUSTERS"))
logger.debug("\n" + "\n".join([str(c) for c in clusters]))
logger.debug("#" * 80)
logger.info("Searching systems")
if model.multi_loci:
# The loners are already in clusters lists with their context
# so they are take in account
clusters_combination = [
itertools.combinations(clusters, i)
for i in range(1,
len(clusters) + 1)
]
else:
# we must add loners manually
# but only if the cluster does not already contains them
loners = cluster.get_loners(hits_related_one_model, model,
hit_weights)
clusters_combination = []
for one_cluster in clusters:
one_clust_combination = [one_cluster]
filtered_loners = cluster.filter_loners(
one_cluster, loners)
one_clust_combination.extend(filtered_loners)
clusters_combination.append([one_clust_combination])
for one_combination_set in clusters_combination:
for one_clust_combination in one_combination_set:
ordered_matcher = OrderedMatchMaker(
model, redundancy_penalty=config.redundancy_penalty())
res = ordered_matcher.match(one_clust_combination)
if isinstance(res, System):
systems.append(res)
else:
rejected_clusters.append(res)
if systems:
systems.sort(key=lambda syst: (syst.replicon_name, syst.position[0],
syst.model.fqn, -syst.score))
return systems, rejected_clusters
def test_fill_gembase_min_max_oredered_replicon(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()
RepliconDB.__init__ = self.fake_init
db = RepliconDB(cfg)
with self.assertRaises(MacsypyError) as ctx:
with self.catch_log() as log:
db._fill_gembase_min_max({}, self.cfg.replicon_topology())
self.assertEqual(
str(ctx.exception),
f"Error during sequence-db '{self.args.sequence_db}' parsing. "
f"Are you sure db-type is 'gembase'?")
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)
self.assertEqual(ESCO030p01, db['ESCO030p01'])
self.assertEqual(PSAE001c01, db['PSAE001c01'])
self.assertRaises(KeyError, db.__getitem__, 'foo')
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_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()
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_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_in(self):
db = RepliconDB(self.cfg)
self.assertIn('ESCO030p01', db)
self.assertIn('PSAE001c01', db)
self.assertNotIn('toto', db)
def test_names(self):
db = RepliconDB(self.cfg)
exp_name = ['ESCO030p01', 'PSAE001c01', 'NC_xxxxx_xx']
self.assertListEqual(db.replicon_names(), exp_name)