def setUp(self):
self.hit_a1 = {
'locus_tag': 'locus_tag_a',
'location': '[1000:2000]',
'label': 'a1_generic_label',
'domain': 'RRE_type_A',
'evalue': 0.1,
'protein_start': 0,
'protein_end': 50,
'score': 38.0,
'description': 'description_type_A',
'translation': 'FAKESEQ',
'identifier': 'RREFam001',
}
self.hit_b1 = {
'locus_tag': 'locus_tag_b',
'location': '[500:1500]',
'label': 'b1_generic_label',
'domain': 'RRE_type_B',
'evalue': 0.2,
'protein_start': 0,
'protein_end': 90,
'score': 25.0,
'description': 'description_type_B',
'translation': 'FAKESEQ',
'identifier': 'RREFam002',
}
self.hit_info = dict(
(hit['locus_tag'], [hit]) for hit in [self.hit_a1, self.hit_b1])
self.hits_per_protocluster = {1: ['locus_tag_a'], 2: ['locus_tag_b']}
self.tool = 'rrefinder'
self.database = 'RREFam.hmm'
self.detection = 'hmmscan'
self.min_length = 50
self.bitscore_cutoff = 25.0
self.record = unittest.mock.Mock()
self.record.id = self.record_id = 'test_record'
self.record2 = unittest.mock.Mock()
self.record2.record_id = 'another_record'
self.json_dict = {
"schema_version": RREFinderResults.schema_version,
"bitscore_cutoff": self.bitscore_cutoff,
"hits_per_protocluster": self.hits_per_protocluster,
"hit_info": self.hit_info,
"min_length": self.min_length,
"record_id": self.record_id,
}
self.res_object = RREFinderResults(self.record_id,
self.bitscore_cutoff,
self.min_length,
self.hits_per_protocluster,
self.hit_info)
def test_from_json(self):
res_object = RREFinderResults.from_json(self.json_dict, self.record,
self.min_length,
self.bitscore_cutoff)
assert res_object
assert res_object.hits_per_protocluster == self.hits_per_protocluster
assert res_object.hit_info == self.hit_info
assert res_object.bitscore_cutoff == self.bitscore_cutoff
assert res_object.min_length == self.min_length
assert res_object.record_id == self.record_id
json_dict2 = dict(self.json_dict)
json_dict2['record_id'] = 'another_record'
assert not RREFinderResults.from_json(
json_dict2, self.record, self.min_length, self.bitscore_cutoff)
json_dict3 = dict(self.json_dict)
json_dict3['schema_version'] = 'not_a_valid_schema'
assert not RREFinderResults.from_json(
json_dict3, self.record, self.min_length, self.bitscore_cutoff)
# Invalid arguments should return None
json_dict4 = dict(self.json_dict)
_ = json_dict4.pop('bitscore_cutoff')
with self.assertRaises(ValueError):
_ = RREFinderResults.from_json(json_dict4, self.record,
self.min_length,
self.bitscore_cutoff)
json_dict5 = dict(self.json_dict)
_ = json_dict5.pop('min_length')
with self.assertRaises(ValueError):
_ = RREFinderResults.from_json(json_dict5, self.record,
self.min_length,
self.bitscore_cutoff)
# More lenient settings shouldn't return results
assert not RREFinderResults.from_json(self.json_dict, self.record, 25,
self.bitscore_cutoff)
assert not RREFinderResults.from_json(self.json_dict, self.record,
self.min_length, 15.0)
# Stricter settings should filter
res_object2 = RREFinderResults.from_json(self.json_dict, self.record,
self.min_length, 35.0)
assert res_object2.hit_info['locus_tag_a'] == [self.hit_a1]
assert len(res_object2.hit_info) == 1
assert res_object2.hits_per_protocluster[1] == ['locus_tag_a']
assert len(res_object2.hits_per_protocluster) == 1
res_object3 = RREFinderResults.from_json(self.json_dict, self.record,
80, self.bitscore_cutoff)
assert res_object3.hits_per_protocluster[2] == ['locus_tag_b']
assert len(res_object3.hits_per_protocluster) == 1
assert res_object3.hit_info['locus_tag_b'] == [self.hit_b1]
assert len(res_object3.hit_info) == 1
def test_json_conversion(self):
results = self.create_results()
json = jsonlib.loads(jsonlib.dumps(results.to_json()))
regenerated = RREFinderResults.from_json(json, self.record)
assert regenerated.hits_by_protocluster == results.hits_by_protocluster
assert regenerated.hits_by_cds == results.hits_by_cds
assert regenerated.bitscore_cutoff == results.bitscore_cutoff
assert regenerated.min_length == results.min_length
assert regenerated.record_id == results.record_id
def test_from_json_higher_bitscore(self):
json = self.create_results().to_json()
assert get_config().rre_cutoff == 25.
new = 35.
assert self.hits[0].score > new
assert self.hits[1].score < new
update_config({"rre_cutoff": new})
result = RREFinderResults.from_json(json, self.record)
assert len(result.hits_by_cds) == 1
assert result.hits_by_cds[self.hits[0].locus_tag] == [self.hits[0]]
assert len(result.hits_by_protocluster) == 1
assert result.hits_by_protocluster[1] == [self.hits[0].locus_tag]
def test_from_json_higher_min_length(self):
json = self.create_results().to_json()
assert get_config().rre_min_length == 50
new = 80
assert len(self.hits[0]) < new
assert len(self.hits[1]) > new
update_config({"rre_min_length": new})
results = RREFinderResults.from_json(json, self.record)
assert len(results.hits_by_cds) == 1
assert results.hits_by_cds[self.hits[1].locus_tag] == [self.hits[1]]
assert len(results.hits_by_protocluster) == 1
assert results.hits_by_protocluster[2] == [self.hits[1].locus_tag]
def create_results(self,
record_id=None,
cutoff=None,
min_length=None,
hits_by_protocluster=None,
hits_by_cds=None):
record_id = record_id or self.record.id
bitscore_cutoff = cutoff if cutoff is not None else self.bitscore_cutoff
min_length = min_length or self.min_length
hits_by_protocluster = hits_by_protocluster or self.hits_by_protocluster
hits_by_cds = hits_by_cds or self.hits_by_cds
return RREFinderResults(record_id, bitscore_cutoff, min_length,
hits_by_protocluster, hits_by_cds)
def test_convert_hits_to_features(self):
hit_info = self.hit_info.copy()
_ = hit_info.pop('locus_tag_b')
res_object = RREFinderResults(self.record_id, self.bitscore_cutoff,
self.min_length,
self.hits_per_protocluster, hit_info)
assert len(res_object.features) == 1
feature = res_object.features[0]
assert isinstance(feature, RRE)
assert feature.location == FeatureLocation(1000, 2000)
assert feature.protein_location == FeatureLocation(0, 50)
assert feature.description == 'description_type_A'
assert feature.domain == 'RRE_type_A'
assert feature.locus_tag == 'locus_tag_a'
assert feature.domain_id == '{}_locus_tag_a_0001'.format(self.tool)
assert feature.database == self.database
assert feature.detection == self.detection
assert feature.score == self.hit_a1['score']
assert feature.evalue == self.hit_a1['evalue']
assert feature.label == self.hit_a1['label']
assert feature.translation == self.hit_a1['translation']
def test_from_json_lower_bitscore(self):
json = self.create_results().to_json()
assert get_config().rre_cutoff == 25.
update_config({"rre_cutoff": 15.})
assert not RREFinderResults.from_json(json, self.record)
def test_from_json_lower_min_length(self):
json = self.create_results().to_json()
assert get_config().rre_min_length == 50
update_config({"rre_min_length": 25})
assert not RREFinderResults.from_json(json, self.record)
def test_from_json_wrong_schema_version(self):
json = self.create_results().to_json()
json['schema_version'] = 'not_a_valid_schema'
assert not RREFinderResults.from_json(json, self.record)
def test_from_json_wrong_record(self):
json = self.create_results().to_json()
json['record_id'] = 'another_record'
assert not RREFinderResults.from_json(json, self.record)
class TestRREResults(unittest.TestCase):
def setUp(self):
self.hit_a1 = {
'locus_tag': 'locus_tag_a',
'location': '[1000:2000]',
'label': 'a1_generic_label',
'domain': 'RRE_type_A',
'evalue': 0.1,
'protein_start': 0,
'protein_end': 50,
'score': 38.0,
'description': 'description_type_A',
'translation': 'FAKESEQ',
'identifier': 'RREFam001',
}
self.hit_b1 = {
'locus_tag': 'locus_tag_b',
'location': '[500:1500]',
'label': 'b1_generic_label',
'domain': 'RRE_type_B',
'evalue': 0.2,
'protein_start': 0,
'protein_end': 90,
'score': 25.0,
'description': 'description_type_B',
'translation': 'FAKESEQ',
'identifier': 'RREFam002',
}
self.hit_info = dict(
(hit['locus_tag'], [hit]) for hit in [self.hit_a1, self.hit_b1])
self.hits_per_protocluster = {1: ['locus_tag_a'], 2: ['locus_tag_b']}
self.tool = 'rrefinder'
self.database = 'RREFam.hmm'
self.detection = 'hmmscan'
self.min_length = 50
self.bitscore_cutoff = 25.0
self.record = unittest.mock.Mock()
self.record.id = self.record_id = 'test_record'
self.record2 = unittest.mock.Mock()
self.record2.record_id = 'another_record'
self.json_dict = {
"schema_version": RREFinderResults.schema_version,
"bitscore_cutoff": self.bitscore_cutoff,
"hits_per_protocluster": self.hits_per_protocluster,
"hit_info": self.hit_info,
"min_length": self.min_length,
"record_id": self.record_id,
}
self.res_object = RREFinderResults(self.record_id,
self.bitscore_cutoff,
self.min_length,
self.hits_per_protocluster,
self.hit_info)
def test_init(self):
assert self.res_object.record_id == self.record_id
assert self.res_object.bitscore_cutoff == self.bitscore_cutoff
assert self.res_object.min_length == self.min_length
assert self.res_object.hits_per_protocluster == self.hits_per_protocluster
assert self.res_object.hit_info == self.hit_info
assert self.res_object.tool == self.tool
assert self.res_object.detection == self.detection
assert self.res_object.database == self.database
assert self.res_object.features
def test_convert_hits_to_features(self):
hit_info = self.hit_info.copy()
_ = hit_info.pop('locus_tag_b')
res_object = RREFinderResults(self.record_id, self.bitscore_cutoff,
self.min_length,
self.hits_per_protocluster, hit_info)
assert len(res_object.features) == 1
feature = res_object.features[0]
assert isinstance(feature, RRE)
assert feature.location == FeatureLocation(1000, 2000)
assert feature.protein_location == FeatureLocation(0, 50)
assert feature.description == 'description_type_A'
assert feature.domain == 'RRE_type_A'
assert feature.locus_tag == 'locus_tag_a'
assert feature.domain_id == '{}_locus_tag_a_0001'.format(self.tool)
assert feature.database == self.database
assert feature.detection == self.detection
assert feature.score == self.hit_a1['score']
assert feature.evalue == self.hit_a1['evalue']
assert feature.label == self.hit_a1['label']
assert feature.translation == self.hit_a1['translation']
def test_to_json(self):
assert self.json_dict == self.res_object.to_json()
def test_from_json(self):
res_object = RREFinderResults.from_json(self.json_dict, self.record,
self.min_length,
self.bitscore_cutoff)
assert res_object
assert res_object.hits_per_protocluster == self.hits_per_protocluster
assert res_object.hit_info == self.hit_info
assert res_object.bitscore_cutoff == self.bitscore_cutoff
assert res_object.min_length == self.min_length
assert res_object.record_id == self.record_id
json_dict2 = dict(self.json_dict)
json_dict2['record_id'] = 'another_record'
assert not RREFinderResults.from_json(
json_dict2, self.record, self.min_length, self.bitscore_cutoff)
json_dict3 = dict(self.json_dict)
json_dict3['schema_version'] = 'not_a_valid_schema'
assert not RREFinderResults.from_json(
json_dict3, self.record, self.min_length, self.bitscore_cutoff)
# Invalid arguments should return None
json_dict4 = dict(self.json_dict)
_ = json_dict4.pop('bitscore_cutoff')
with self.assertRaises(ValueError):
_ = RREFinderResults.from_json(json_dict4, self.record,
self.min_length,
self.bitscore_cutoff)
json_dict5 = dict(self.json_dict)
_ = json_dict5.pop('min_length')
with self.assertRaises(ValueError):
_ = RREFinderResults.from_json(json_dict5, self.record,
self.min_length,
self.bitscore_cutoff)
# More lenient settings shouldn't return results
assert not RREFinderResults.from_json(self.json_dict, self.record, 25,
self.bitscore_cutoff)
assert not RREFinderResults.from_json(self.json_dict, self.record,
self.min_length, 15.0)
# Stricter settings should filter
res_object2 = RREFinderResults.from_json(self.json_dict, self.record,
self.min_length, 35.0)
assert res_object2.hit_info['locus_tag_a'] == [self.hit_a1]
assert len(res_object2.hit_info) == 1
assert res_object2.hits_per_protocluster[1] == ['locus_tag_a']
assert len(res_object2.hits_per_protocluster) == 1
res_object3 = RREFinderResults.from_json(self.json_dict, self.record,
80, self.bitscore_cutoff)
assert res_object3.hits_per_protocluster[2] == ['locus_tag_b']
assert len(res_object3.hits_per_protocluster) == 1
assert res_object3.hit_info['locus_tag_b'] == [self.hit_b1]
assert len(res_object3.hit_info) == 1
def test_add_record(self):
with self.assertRaisesRegex(
ValueError,
"Record to store in and record analysed don't match"):
self.res_object.add_to_record(self.record2)
self.res_object.add_to_record(self.record)
assert self.record.add_feature.call_count == 2
for arg in self.record.add_feature.call_args_list:
# First argument of positional arguments
assert isinstance(arg[0][0], RRE)