def test_blank_records(self):
blank_no_pfams = DummyRecord()
blank_no_ids = Record(Seq("ATGTTATGAGGGTCATAACAT", generic_dna))
fake_pfam = DummyPFAMDomain(identifier="PF00000")
blank_no_ids.add_pfam_domain(fake_pfam)
assert not pfam2go.get_gos_for_pfams(blank_no_pfams)
assert not pfam2go.get_gos_for_pfams(blank_no_ids)
def test_get_gos_id_handling(self):
pfams = {
'PF00015.42': FeatureLocation(0, 3),
'PF0015.42': FeatureLocation(0, 3)
}
fake_record = set_dummy_with_pfams(pfams)
with self.assertRaisesRegex(ValueError,
'Pfam id PF0015 is not a valid Pfam id'):
pfam2go.get_gos_for_pfams(fake_record)
def test_blank_records(self):
blank_no_pfams = DummyRecord()
blank_no_ids = Record(Seq("ATGTTATGAGGGTCATAACAT", generic_dna))
fake_pfam_location = FeatureLocation(0, 12)
fake_pfam = PFAMDomain(location=fake_pfam_location,
description='MCPsignal',
protein_start=0,
protein_end=5,
identifier="PF00000",
tool="test")
fake_pfam.domain_id = 'BLANK'
blank_no_ids.add_pfam_domain(fake_pfam)
assert not pfam2go.get_gos_for_pfams(blank_no_pfams)
assert not pfam2go.get_gos_for_pfams(blank_no_ids)
def test_to_json(self):
fake_pfam_location = FeatureLocation(0, 12)
pfams = {'PF00015': fake_pfam_location, 'PF00351': fake_pfam_location}
fake_record = set_dummy_with_pfams(pfams)
gos_for_fake_pfam = pfam2go.get_gos_for_pfams(fake_record)
fake_results = pfam2go.Pfam2GoResults(fake_record.id,
gos_for_fake_pfam)
result_json = fake_results.to_json()
expected_result = {
"pfams": {
"PF00015": {
"GO:0004871": "signal transducer activity",
"GO:0007165": "signal transduction",
"GO:0016020": "membrane"
},
"PF00351": {
"GO:0016714":
("oxidoreductase activity, acting on paired donors, "
"with incorporation or reduction of molecular oxygen, "
"reduced pteridine as one donor, and incorporation of "
"one atom of oxygen"),
"GO:0055114":
"oxidation-reduction process"
}
},
"record_id": fake_record.id,
"schema_version": 1
}
assert result_json["record_id"] == expected_result["record_id"]
assert result_json["schema_version"] == 1
for pfam in expected_result["pfams"]:
assert expected_result["pfams"][pfam] == result_json["pfams"][pfam]
def test_add_results_to_record(self):
pfams = {
'PF00015.2': FeatureLocation(0, 3),
'PF00351.1': FeatureLocation(0, 3),
'PF00015.27': FeatureLocation(3, 6)
}
fake_record = set_dummy_with_pfams(pfams)
fake_duplicate_pfam = PFAMDomain(location=FeatureLocation(6, 9),
description='DUPLICATE',
protein_start=0,
protein_end=5,
identifier="PF00015.2",
tool="test")
fake_duplicate_pfam.domain_id = 'DUPLICATE'
fake_record.add_pfam_domain(fake_duplicate_pfam)
assert fake_duplicate_pfam in fake_record.get_pfam_domains()
gos_for_fake_pfam = pfam2go.get_gos_for_pfams(fake_record)
fake_results = pfam2go.Pfam2GoResults(fake_record.id,
gos_for_fake_pfam)
fake_results.add_to_record(fake_record)
assert fake_duplicate_pfam.full_identifier == 'PF00015.2'
for pfam in fake_record.get_pfam_domains():
assert sorted(pfam.gene_ontologies.ids) == sorted(
fake_results.get_all_gos(pfam))
# make sure identical pfams (with different version numbers) all have the same gene ontologies
if pfam.identifier == "PF00015":
assert pfam.version in [2, 27]
assert sorted(pfam.gene_ontologies.ids) == sorted(
fake_results.get_all_gos(fake_duplicate_pfam))
def test_blank_records(self):
blank_no_pfams = DummyRecord()
blank_no_ids = Record(Seq("ATGTTATGAGGGTCATAACAT", generic_dna))
fake_pfam_location = FeatureLocation(0, 12)
fake_pfam = PFAMDomain(location=fake_pfam_location,
description='MCPsignal',
protein_start=0,
protein_end=5)
fake_pfam.domain_id = 'BLANK'
blank_no_ids.add_pfam_domain(fake_pfam)
with self.assertLogs(level='DEBUG') as log_cm:
gos_for_no_pfams = pfam2go.get_gos_for_pfams(blank_no_pfams)
assert 'No Pfam domains found' in str(log_cm.output)
assert not gos_for_no_pfams
gos_for_no_ids = pfam2go.get_gos_for_pfams(blank_no_ids)
assert 'No Pfam ids found' in str(log_cm.output)
assert not gos_for_no_ids
def test_adding_to_wrong_record(self):
pfams = {'PF00015': FeatureLocation(0, 3)}
fake_record = set_dummy_with_pfams(pfams)
gos_for_fake_pfam = pfam2go.get_gos_for_pfams(fake_record)
fake_results = pfam2go.Pfam2GoResults('FAKEID', gos_for_fake_pfam)
with self.assertRaisesRegex(
ValueError,
"Record to store in and record analysed don't match"):
fake_results.add_to_record(fake_record)
def test_results(self):
pfams = {'PF00015': FeatureLocation(0, 3)}
fake_record = set_dummy_with_pfams(pfams)
gos_for_fake_pfam = pfam2go.get_gos_for_pfams(fake_record)
fake_results = pfam2go.Pfam2GoResults(fake_record.id,
gos_for_fake_pfam)
assert gos_for_fake_pfam == fake_results.pfam_domains_with_gos
assert fake_record.id == fake_results.record_id
for pfam, all_ontologies in fake_results.pfam_domains_with_gos.items():
for ontologies in all_ontologies:
assert ontologies.pfam == pfam.identifier
def test_get_gos(self):
pfams = {
'PF00015': FeatureLocation(0, 3),
'PF00351.42': FeatureLocation(6, 12)
}
fake_record = set_dummy_with_pfams(pfams)
gos_for_fake_pfam = pfam2go.get_gos_for_pfams(fake_record)
for all_ontologies in gos_for_fake_pfam.values():
for ontologies in all_ontologies:
go_ids = [str(go_entry) for go_entry in ontologies.go_entries]
for go_id in go_ids:
assert go_id in self.known_connections[ontologies.pfam]
def test_results(self):
pfams = {'PF00015': FeatureLocation(0, 3)}
fake_record = set_dummy_with_pfams(pfams)
gos_for_fake_pfam = pfam2go.get_gos_for_pfams(fake_record)
fake_results = pfam2go.Pfam2GoResults(fake_record.id,
gos_for_fake_pfam)
assert gos_for_fake_pfam == fake_results.pfam_domains_with_gos
assert fake_record.id == fake_results.record_id
for pfam, all_ontologies in fake_results.pfam_domains_with_gos.items():
pfam_ids_without_versions = [
pfam_id.partition('.')[0] for pfam_id in pfam.db_xref
]
for ontologies in all_ontologies:
assert ontologies.pfam in pfam_ids_without_versions
def test_from_json(self):
fake_pfam_location = FeatureLocation(0, 12)
pfams = {
'PF00015': fake_pfam_location,
'PF00351': fake_pfam_location,
'PF05147': fake_pfam_location
}
fake_record = set_dummy_with_pfams(pfams)
gos_for_fake_pfam = pfam2go.get_gos_for_pfams(fake_record)
fake_results = pfam2go.Pfam2GoResults(fake_record.id,
gos_for_fake_pfam)
result_json = fake_results.to_json()
results_from_json = pfam2go.Pfam2GoResults.from_json(
result_json, fake_record)
assert 'PF05147' not in result_json["pfams"]
for pfam in results_from_json.pfam_domains_with_gos:
assert pfam.identifier in result_json["pfams"]
from_json_to_json = results_from_json.to_json()
assert result_json == from_json_to_json
assert from_json_to_json["schema_version"] == 1
