def test_add_forbidden_gene(self):
system = System(self.cfg, "foo", 10)
gene = Gene(self.cfg, 'sctJ_FLG', system, self.profile_registry)
system.add_forbidden_gene(gene)
self.assertEqual(system._forbidden_genes, [gene])
self.assertEqual(system._accessory_genes, [])
self.assertEqual(system._mandatory_genes, [])
def test_add_forbidden_gene(self):
system = System(self.cfg, "foo", 10)
gene = Gene(self.cfg, 'sctJ_FLG', system, self.profile_registry)
system.add_forbidden_gene(gene)
self.assertEqual(system._forbidden_genes, [gene])
self.assertEqual(system._accessory_genes, [])
self.assertEqual(system._mandatory_genes, [])
def test_multi_loci(self):
name = 'True'
inter_gene_max_space = 40
system = System(self.cfg, name, inter_gene_max_space, multi_loci=True)
self.assertTrue(system.multi_loci)
name = 'False'
inter_gene_max_space = 40
system = System(self.cfg, name, inter_gene_max_space)
self.assertFalse(system.multi_loci)
def test_iter(self):
systems = [System(self.cfg, 'foo', 10), System(self.cfg, 'bar', 10)]
for s in systems:
system_bank.add_system(s)
i = 0
for s in system_bank:
self.assertIn(s, systems)
i = i + 1
self.assertEqual(i, len(systems))
def test_add_homolog(self):
system_foo = System(self.cfg, "foo", 10)
system_bar = System(self.cfg, "bar", 10)
gene = Gene(self.cfg, 'sctJ_FLG', system_foo, self.profile_registry)
gene_ref = Gene(self.cfg, 'sctJ', system_bar, self.profile_registry)
homolog = Homolog(self.cfg, gene, gene_ref)
gene.add_homolog(homolog)
self.assertEqual(len(gene.homologs), 1)
self.assertEqual(gene.homologs[0], homolog)
def test_min_mandatory_genes_required(self):
name = 'foo'
min_mandatory_genes_required = 40
system = System(self.cfg, name, 10, min_mandatory_genes_required = min_mandatory_genes_required)
gene = Gene(self.cfg, 'sctJ_FLG', system, self.profile_registry)
system.add_mandatory_gene( gene )
self.assertEqual(system.min_mandatory_genes_required, min_mandatory_genes_required)
#see https://projets.pasteur.fr/issues/1850
system = System(self.cfg, name, 10)
self.assertEqual(system.min_mandatory_genes_required, len(system.mandatory_genes))
def test_get_homologs(self):
system_foo = System(self.cfg, "foo", 10)
system_bar = System(self.cfg, "bar", 10)
gene = Gene(self.cfg, 'sctN', system_foo, self.profile_registry)
sctJ_FLG = Gene(self.cfg, 'sctJ_FLG', system_foo,
self.profile_registry)
sctJ = Gene(self.cfg, 'sctJ', system_bar, self.profile_registry)
homolog_1 = Homolog(sctJ_FLG, gene)
gene.add_homolog(homolog_1)
homolog_2 = Homolog(sctJ, gene)
gene.add_homolog(homolog_2)
self.assertEqual(gene.get_homologs(), [homolog_1, homolog_2])
def test_max_nb_genes(self):
name = 'foo'
inter_gene_max_space = 40
max_nb_genes = 10
system = System(self.cfg,
name,
inter_gene_max_space,
max_nb_genes=max_nb_genes)
self.assertEqual(system.max_nb_genes, max_nb_genes)
name = 'bar'
system = System(self.cfg, name, inter_gene_max_space)
self.assertIsNone(system.max_nb_genes)
def test_str(self):
"""
"""
system_foo = System(self.cfg, "foo", 10)
gene = Gene(self.cfg, 'sctJ_FLG', system_foo, self.profile_registry)
system_bar = System(self.cfg, "bar", 20)
gene_homolog = Gene(self.cfg, 'sctJ', system_bar,
self.profile_registry)
homolog = Homolog(gene_homolog, gene, self.cfg)
gene.add_homolog(homolog)
s = """name : sctJ_FLG
inter_gene_max_space: 10
homologs: sctJ"""
self.assertEqual(str(gene), s)
def test_contains(self):
system_foo = System(self.cfg, "foo", 10)
gene_in = Gene(self.cfg, 'sctJ_FLG', system_foo, self.profile_registry)
gene_bank.add_gene(gene_in)
self.assertIn(gene_in, gene_bank)
gene_out = Gene(self.cfg, 'abc', system_foo, self.profile_registry)
self.assertNotIn( gene_out, gene_bank)
def test_add_get_system(self):
system_name = 'foo'
self.assertRaises(KeyError, system_bank.__getitem__, system_name)
system_foo = System(self.cfg, system_name, 10)
system_bank.add_system(system_foo)
self.assertTrue(isinstance(system_foo, System))
self.assertEqual(system_foo, system_bank[system_name])
def test_get_uniq_object(self):
system_foo = System(self.cfg, "foo", 10)
gene = Gene(self.cfg, 'sctJ_FLG', system_foo, self.profile_registry)
path = self.profile_registry.get('sctJ_FLG')
profile1 = profile_factory.get_profile(gene, self.cfg, path)
profile2 = profile_factory.get_profile(gene, self.cfg, path)
self.assertEqual(profile1, profile2)
def test_get_profile(self):
system_foo = System(self.cfg, "foo", 10)
gene_name = 'sctJ_FLG'
gene = Gene(self.cfg, gene_name, system_foo, self.profile_registry)
profile = profile_factory.get_profile(gene, self.cfg, self.profile_registry )
self.assertTrue( isinstance( profile, Profile ))
self.assertEqual( profile.gene.name, gene_name )
def test_str(self):
system = System(self.cfg, "T2SS", 10)
gene_name = "gspD"
gene = Gene(self.cfg, "gspD", system, self.profile_registry)
hit_prop = {
'id': "PSAE001c01_006940",
'hit_seq_len': 803,
'replicon_name': "PSAE001c01",
'position': 694,
'i_eval': float(1.2e-234),
'score': float(779.2),
'gene_name': gene.name,
'system_name': system.name,
'profil_coverage': float(1.0),
'sequence_coverage': float(638.000000),
'begin': 104,
'end': 741
}
hit = Hit(gene, system, hit_prop['id'], hit_prop['hit_seq_len'],
hit_prop['replicon_name'], hit_prop['position'],
hit_prop['i_eval'], hit_prop['score'],
hit_prop['profil_coverage'], hit_prop['sequence_coverage'],
hit_prop['begin'], hit_prop['end'])
s = "%(id)s\t%(replicon_name)s\t%(position)d\t%(hit_seq_len)d\t%(gene_name)s\t%(system_name)s\t%(i_eval)s\t%(score)s\t%(profil_coverage)f\t%(sequence_coverage)f\t%(begin)d\t%(end)d\n" % hit_prop
self.assertEqual(s, str(hit))
def test_str(self):
system = System(self.cfg, "T2SS", 10)
gene_name = "gspD"
gene = Gene(self.cfg, "gspD", system, self.profile_registry)
hit_prop = {
'id': "PSAE001c01_006940",
'hit_seq_len': 803,
'replicon_name': "PSAE001c01",
'position': 694,
'i_eval': float(1.2e-234),
'score': float(779.2),
'gene_name': gene.name,
'system_name': system.name,
'profil_coverage': float(1.0),
'sequence_coverage': float(638.000000),
'begin': 104,
'end': 741
}
hit = Hit(gene, system, hit_prop['id'], hit_prop['hit_seq_len'],
hit_prop['replicon_name'], hit_prop['position'],
hit_prop['i_eval'], hit_prop['score'],
hit_prop['profil_coverage'], hit_prop['sequence_coverage'],
hit_prop['begin'], hit_prop['end'])
s = "{id}\t{replicon_name}\t{position:d}\t{hit_seq_len:d}\t{gene_name}\t{system_name}\t{i_eval:.3e}\t{score:.3f}\t{profil_coverage:.3f}\t{sequence_coverage:.3f}\t{begin:d}\t{end:d}\n".format(
**hit_prop)
self.assertEqual(s, str(hit))
def test_str(self):
system = System(self.cfg, "T2SS", 10)
gene = Gene(self.cfg, "abc", system, self.profile_registry)
path = self.profile_registry.get("abc")
profile = Profile(gene, self.cfg, path)
s = "{0} : {1}".format(gene.name, path)
self.assertEqual(str(profile), s)
def test_execute_unknown_binary(self):
self.cfg.options['hmmer_exe'] = "Nimportnaoik"
system = System(self.cfg, "T2SS", 10)
gene = Gene(self.cfg, "abc", system, self.profile_registry)
path = self.profile_registry.get("abc")
profile = Profile(gene, self.cfg, path)
self.assertRaises(RuntimeError, profile.execute)
def test_GembaseHMMReport_extract(self):
system = System(self.cfg, "T2SS", 10)
gene_name = "gspD"
gene = Gene(self.cfg, gene_name, system, self.profile_registry)
shutil.copy(
os.path.join(self._data_dir, "hmm",
gene_name + self.cfg.res_search_suffix),
self.cfg.working_dir)
report_path = os.path.join(self.cfg.working_dir,
gene_name + self.cfg.res_search_suffix)
report = GembaseHMMReport(gene, report_path, self.cfg)
report.extract()
self.assertEqual(len(report.hits), 6)
# gene, system, hit_id, hit_seq_length replicon_name, pos_hit, i_eval, score, profile_coverage, sequence_coverage, begin_match, end_match
hits = [
Hit(gene, system, "NC_xxxxx_xx_056141", 803, "NC_xxxxx_xx", 141,
float(2e-236), float(779.2), float(1.000000),
(741.0 - 104.0 + 1) / 803, 104, 741),
Hit(gene, system, "PSAE001c01_006940", 803, "PSAE001c01", 68,
float(1.2e-234), float(779.2), float(1.000000),
(741.0 - 104.0 + 1) / 803, 104, 741),
Hit(gene, system, "PSAE001c01_013980", 759, "PSAE001c01", 69,
float(3.7e-76), float(255.8), float(1.000000),
(736.0 - 105.0 + 1) / 759, 105, 736),
Hit(gene, system, "PSAE001c01_017350", 600, "PSAE001c01", 70,
float(3.2e-27), float(94.2), float(0.500000),
(506.0 - 226.0 + 1) / 600, 226, 506),
Hit(gene, system, "PSAE001c01_018920", 776, "PSAE001c01", 71,
float(6.1e-183), float(608.4), float(1.000000),
(606.0 - 48.0 + 1) / 776, 48, 606),
Hit(gene, system, "PSAE001c01_031420", 658, "PSAE001c01", 73,
float(1.8e-210), float(699.3), float(1.000000),
(614.0 - 55.0 + 1) / 658, 55, 614)
]
self.assertListEqual(hits, report.hits)
def test_system(self):
"""
test getter/setter for system property
"""
system_foo = System(self.cfg, "foo", 10)
gene = Gene(self.cfg, 'sctJ_FLG', system_foo, self.profile_registry)
self.assertEqual(gene.system, system_foo)
def test_get_uniq_object(self):
system_foo = System(self.cfg, "foo", 10)
gene_in = Gene(self.cfg, 'sctJ_FLG', system_foo, self.profile_registry)
gene_bank.add_gene(gene_in)
gene1 = gene_bank['sctJ_FLG']
gene2 = gene_bank['sctJ_FLG']
self.assertEqual(gene1, gene2)
self.assertIs(gene1, gene2)
def test_is_aligned(self):
system = System(self.cfg, "T2SS", 10)
gene_ref = Gene(self.cfg, 'sctJ_FLG', system, self.profile_registry)
gene = Gene(self.cfg, 'sctJ', system, self.profile_registry)
homolog = Homolog(gene, gene_ref)
self.assertFalse(homolog.is_aligned())
homolog = Homolog(gene, gene_ref, aligned=True)
self.assertTrue(homolog.is_aligned())
def test_add_get_gene(self):
gene_name = 'sctJ_FLG'
self.assertRaises(KeyError, gene_bank.__getitem__, gene_name)
system_foo = System( "foo", self.cfg, 10)
gene = Gene(self.cfg, gene_name, system_foo, self.profile_registry)
gene_bank.add_gene(gene)
gene_from_bank = gene_bank[gene_name]
self.assertTrue(isinstance(gene_from_bank, Gene))
self.assertEqual(gene_from_bank, gene)
def test_SystemSerializer_tsv(self):
model = Model("foo/T2SS", 10)
c_gene_gspd = CoreGene(self.model_location, "gspD",
self.profile_factory)
gene_gspd = ModelGene(c_gene_gspd, model)
model.add_mandatory_gene(gene_gspd)
c_gene_sctj = CoreGene(self.model_location, "sctJ",
self.profile_factory)
gene_sctj = ModelGene(c_gene_sctj, model)
model.add_accessory_gene(gene_sctj)
c_gene_sctn = CoreGene(self.model_location, "sctN",
self.profile_factory)
gene_sctn = ModelGene(c_gene_sctn, model)
c_gene_sctn_flg = CoreGene(self.model_location, "sctN_FLG",
self.profile_factory)
gene_sctn_flg = Exchangeable(c_gene_sctn_flg, gene_sctn)
gene_sctn.add_exchangeable(gene_sctn_flg)
model.add_accessory_gene(gene_sctn)
h_gspd = Hit(c_gene_gspd, "h_gspd", 803, "replicon_id", 10, 1.0, 1.0,
1.0, 1.0, 10, 20)
v_h_gspd = ValidHit(h_gspd, gene_gspd, GeneStatus.MANDATORY)
h_sctj = Hit(c_gene_sctj, "h_sctj", 803, "replicon_id", 20, 1.0, 1.0,
1.0, 1.0, 20, 30)
v_h_sctj = ValidHit(h_sctj, gene_sctj, GeneStatus.ACCESSORY)
h_sctn_flg = Hit(c_gene_sctn_flg, "h_sctn_flg", 803, "replicon_id", 30,
1.0, 1.0, 1.0, 1.0, 30, 40)
v_h_sctn_flg = ValidHit(h_sctn_flg, gene_sctn_flg,
GeneStatus.ACCESSORY)
c1 = Cluster([v_h_gspd, v_h_sctj], model, self.hit_weights)
c2 = Cluster([v_h_sctn_flg], model, self.hit_weights)
sys_multi_loci = System(model, [c1, c2], self.cfg.redundancy_penalty())
hit_multi_sys_tracker = HitSystemTracker([sys_multi_loci])
system_serializer = TsvSystemSerializer()
sys_tsv = "\t".join([
"replicon_id", "h_gspd", "gspD", "10", "foo/T2SS",
sys_multi_loci.id, "1", "1.000", "1.900", "1", "gspD", "mandatory",
"803", "1.0", "1.000", "1.000", "1.000", "10", "20", ""
])
sys_tsv += "\n"
sys_tsv += "\t".join([
"replicon_id", "h_sctj", "sctJ", "20", "foo/T2SS",
sys_multi_loci.id, "1", "1.000", "1.900", "1", "sctJ", "accessory",
"803", "1.0", "1.000", "1.000", "1.000", "20", "30", ""
])
sys_tsv += "\n"
sys_tsv += "\t".join([
"replicon_id", "h_sctn_flg", "sctN_FLG", "30", "foo/T2SS",
sys_multi_loci.id, "1", "1.000", "1.900", "1", "sctN", "accessory",
"803", "1.0", "1.000", "1.000", "1.000", "30", "40", ""
])
sys_tsv += "\n"
self.assertEqual(
sys_tsv,
system_serializer.serialize(sys_multi_loci, hit_multi_sys_tracker))
def test_iter(self):
system_foo = System(self.cfg, "foo", 10)
genes = [Gene(self.cfg, 'sctJ_FLG', system_foo, self.profile_registry), Gene(self.cfg, 'abc', system_foo, self.profile_registry)]
for g in genes:
gene_bank.add_gene(g)
i = 0
for g in gene_bank:
self.assertIn(g, genes)
i = i + 1
self.assertEqual(i, len(genes))
def test_HMMReport(self):
system = System(self.cfg, "T2SS", 10)
gene_name = "gspD"
gene = Gene(self.cfg, gene_name, system, self.profile_registry)
shutil.copy(
os.path.join(self._data_dir, "hmm",
gene_name + self.cfg.res_search_suffix),
self.cfg.working_dir)
report_path = os.path.join(self.cfg.working_dir,
gene_name + self.cfg.res_search_suffix)
self.assertRaises(TypeError, HMMReport, gene, report_path, self.cfg)
def test_GembaseHMMReport_extract_concurent(self):
system = System(self.cfg, "T2SS", 10)
gene_name = "gspD"
gene = Gene(self.cfg, gene_name, system, self.profile_registry)
shutil.copy(
os.path.join(self._data_dir, "hmm",
gene_name + self.cfg.res_search_suffix),
self.cfg.working_dir)
report_path = os.path.join(self.cfg.working_dir,
gene_name + self.cfg.res_search_suffix)
reports = []
for i in range(5):
report = GembaseHMMReport(gene, report_path, self.cfg)
reports.append(report)
import threading
def worker(report):
report.extract()
for report in reports:
t = threading.Thread(target=worker, args=(report, ))
t.start()
main_thread = threading.currentThread()
for t in threading.enumerate():
if t is main_thread:
continue
t.join()
#gene, system, hit_id, hit_seq_length replicon_name, pos_hit, i_eval, score, profile_coverage, sequence_coverage, begin_match, end_match
hits = [
Hit(gene, system, "NC_xxxxx_xx_056141", 803, "NC_xxxxx_xx", 141,
float(2e-236), float(779.2), float(1.000000),
(741.0 - 104.0 + 1) / 803, 104, 741),
Hit(gene, system, "PSAE001c01_006940", 803, "PSAE001c01", 68,
float(1.2e-234), float(779.2), float(1.000000),
(741.0 - 104.0 + 1) / 803, 104, 741),
Hit(gene, system, "PSAE001c01_013980", 759, "PSAE001c01", 69,
float(3.7e-76), float(255.8), float(1.000000),
(736.0 - 105.0 + 1) / 759, 105, 736),
Hit(gene, system, "PSAE001c01_017350", 600, "PSAE001c01", 70,
float(3.2e-27), float(94.2), float(0.500000),
(506.0 - 226.0 + 1) / 600, 226, 506),
Hit(gene, system, "PSAE001c01_018920", 776, "PSAE001c01", 71,
float(6.1e-183), float(608.4), float(1.000000),
(606.0 - 48.0 + 1) / 776, 48, 606),
Hit(gene, system, "PSAE001c01_031420", 658, "PSAE001c01", 73,
float(1.8e-210), float(699.3), float(1.000000),
(614.0 - 55.0 + 1) / 658, 55, 614)
]
for report in reports:
report.save_extract()
self.assertEqual(len(report.hits), len(hits))
self.assertListEqual(report.hits, hits)
def test_cmp(self):
system = System(self.cfg, "T2SS", 10)
gene_name = "gspD"
gene = Gene(self.cfg, "gspD", system, self.profile_registry)
h0 = Hit(gene, system, "PSAE001c01_006940", 803, "PSAE001c01", 3450,
float(1.2e-234), float(779.2), float(1.000000),
(741.0 - 104.0 + 1) / 803, 104, 741)
h1 = Hit(gene, system, "PSAE001c01_013980", 759, "PSAE001c01", 4146,
float(3.7e-76), float(255.8), float(1.000000),
(736.0 - 105.0 + 1) / 759, 105, 736)
self.assertGreater(h1, h0)
self.assertLess(h0, h1)
def test_exchangeable(self):
"""
test getter for exchangeable property
"""
system_foo = System(self.cfg, "foo", 10)
gene = Gene(self.cfg, 'sctJ_FLG', system_foo, self.profile_registry)
self.assertFalse(gene.exchangeable)
gene = Gene(self.cfg,
'sctJ',
system_foo,
self.profile_registry,
exchangeable=True)
self.assertTrue(gene.exchangeable)
def test_multi_system(self):
"""
test getter for multi_system property
"""
system_foo = System(self.cfg, "foo", 10)
gene = Gene(self.cfg, 'sctJ_FLG', system_foo, self.profile_registry)
self.assertFalse(gene.multi_system)
gene = Gene(self.cfg,
'sctJ',
system_foo,
self.profile_registry,
multi_system=True)
self.assertTrue(gene.multi_system)
def test_loner(self):
"""
test getter for loner property
"""
system_foo = System(self.cfg, "foo", 10)
gene = Gene(self.cfg, 'sctJ_FLG', system_foo, self.profile_registry)
self.assertFalse(gene.loner)
gene = Gene(self.cfg,
'sctJ',
system_foo,
self.profile_registry,
loner=True)
self.assertTrue(gene.loner)
def test_inter_gene_max_space(self):
"""
test getter for inter_gene_max_space property
"""
system_inter_gene_max_space = 40
gene_inter_gene_max_space = 50
system_foo = System(self.cfg, "foo", system_inter_gene_max_space)
gene = Gene(self.cfg, 'sctJ_FLG', system_foo, self.profile_registry)
self.assertEqual(gene.inter_gene_max_space,
system_inter_gene_max_space)
gene = Gene(self.cfg,
'sctJ',
system_foo,
self.profile_registry,
inter_gene_max_space=gene_inter_gene_max_space)
self.assertEqual(gene.inter_gene_max_space, gene_inter_gene_max_space)
def test_execute(self):
system = System(self.cfg, "T2SS", 10)
gene = Gene(self.cfg, "abc", system, self.profile_registry)
path = self.profile_registry.get("abc")
profile = Profile(gene, self.cfg, path)
report = profile.execute()
hmmer_raw_out = profile.hmm_raw_output
with open(hmmer_raw_out, 'r') as hmmer_raw_out_file:
first_l = hmmer_raw_out_file.readline()
# a hmmsearch output file has been produced
self.assertTrue(first_l.startswith("# hmmsearch :: search profile(s) against a sequence database"))
for i in range(5):
# skip 4 lines
l = hmmer_raw_out_file.readline()
# a hmmsearch used the abc profile line should become with: "# query HMM file:"
path = os.path.join(self.cfg.profile_dir, gene.name + self.cfg.profile_suffix)
self.assertTrue(l.find(path) != -1)
def test_get_gene_ref(self):
system = System(self.cfg, "foo", 10)
gene_name = 'sctJ_FLG'
gene_ref = Gene(self.cfg, gene_name, system, self.profile_registry)
homolog_name = 'sctJ'
gene_homolg = Gene(self.cfg, homolog_name, system, self.profile_registry)
homolog = Homolog(gene_homolg, gene_ref)
gene_ref.add_homolog(homolog)
for meth in (system.add_forbidden_gene, system.add_accessory_gene, system.add_mandatory_gene):
system._mandatory_genes = []
system._accessory_genes = []
system._forbidden_genes = []
meth(gene_ref)
self.assertEqual(gene_ref, system.get_gene_ref(homolog))
self.assertIsNone(system.get_gene_ref(gene_ref))
gene_ukn = Gene(self.cfg, 'abc', system, self.profile_registry)
self.assertRaises(KeyError, system.get_gene_ref, gene_ukn)
def test_mandatory_genes(self):
system = System(self.cfg, "foo", 10)
gene = Gene(self.cfg, 'sctJ_FLG', system, self.profile_registry)
system.add_mandatory_gene(gene)
self.assertEqual(system.mandatory_genes, [gene])
def test_get_gene(self):
system = System(self.cfg, "foo", 10)
gene_name = 'sctJ_FLG'
gene = Gene(self.cfg, gene_name, system, self.profile_registry)
for meth in (system.add_forbidden_gene, system.add_accessory_gene, system.add_mandatory_gene):
system._mandatory_genes = []
system._accessory_genes = []
system._forbidden_genes = []
meth(gene)
self.assertEqual(gene, system.get_gene(gene_name))
self.assertRaises(KeyError, system.get_gene, 'bar')
homolog_name = 'sctJ'
gene_homolog = Gene(self.cfg, homolog_name, system, self.profile_registry)
homolog = Homolog(gene_homolog, gene)
gene.add_homolog(homolog)
for meth in (system.add_forbidden_gene, system.add_accessory_gene, system.add_mandatory_gene):
system._mandatory_genes = []
system._accessory_genes = []
system._forbidden_genes = []
meth(gene)
self.assertEqual(homolog, system.get_gene(homolog_name))
analog_name = 'sctC'
gene_analog = Gene(self.cfg, analog_name, system, self.profile_registry)
analog = Analog(gene_analog, gene)
gene.add_analog(analog)
for meth in (system.add_forbidden_gene, system.add_accessory_gene, system.add_mandatory_gene):
system._mandatory_genes = []
system._accessory_genes = []
system._forbidden_genes = []
meth(gene)
self.assertEqual(analog, system.get_gene(analog_name))
