def test_construction(self):
strings = refinement.HMMResult("dummy_hit", "1", "5", "3E-10", "53.5")
correct = refinement.HMMResult("dummy_hit", 1, 5, 3e-10, 53.5)
for result in [strings, correct]:
assert result.hit_id == "dummy_hit"
assert result.query_start == 1
assert result.query_end == 5
assert result.evalue == 3e-10
assert result.bitscore == 53.5
def test_merge(self):
first = refinement.HMMResult("dummy_hit", 1, 5, 3e-10, 53.5)
second = refinement.HMMResult("dummy_hit", 15, 25, 3e-20, 73.5)
for merged in [first.merge(second), second.merge(first)]:
assert merged.hit_id == "dummy_hit"
assert merged.query_start == 1
assert merged.query_end == 25
assert merged.evalue == 3e-20
assert merged.bitscore == 73.5
def test_equality(self):
first = refinement.HMMResult("dummy_hit", 1, 5, 3e-10, 53.5)
second = refinement.HMMResult("dummy_hit", 1, 5, 3e-10, 53.5)
assert first == second and first is not second
second._hit_id = "dummy"
assert first != second
second._hit_id = first._hit_id
second._evalue /= 10
assert first != second
def test_hashability(self):
first = refinement.HMMResult("dummy_hit", 1, 5, 3e-10, 53.5)
second = refinement.HMMResult("dummy_hit", 1, 5, 3e-10, 53.5)
different = refinement.HMMResult("dummy_hit", 1, 5, 3e-1, 53.5)
assert hash(first) == hash(second) and first is not second
assert hash(first) != hash(different)
used = {}
used[first] = 1
used[second] = 2
used[different] = 3
assert used == {first: 2, different: 3}
def test_json_conversion(self):
result = refinement.HMMResult("dummy_hit", 1, 5, 3e-10, 53.5)
assert result.to_json() == {
'bitscore': 53.5,
'evalue': 3e-10,
'hit_id': 'dummy_hit',
'query_end': 5,
'query_start': 1
}
def test_overlaps(self):
first = refinement.HMMResult("dummy_hit", 1, 10, 3e-10, 53.5)
second = refinement.HMMResult("dummy_hit", 10, 20, 3e-20, 73.5)
results = [first, second]
assert len(refinement._remove_overlapping(results,
{"dummy_hit": 20})) == 2
assert len(refinement._remove_overlapping(results,
{"dummy_hit": 100})) == 2
first.query_end = 16
assert len(refinement._remove_overlapping(results,
{"dummy_hit": 20})) == 1
assert len(refinement._remove_overlapping(results,
{"dummy_hit": 100})) == 2
first.query_end = 13
assert len(refinement._remove_overlapping(results,
{"dummy_hit": 10})) == 1
assert len(refinement._remove_overlapping(results,
{"dummy_hit": 100})) == 2
def from_json(json, _record) -> "SMCOGResults":
if json.get("schema_version") != SMCOGResults.schema_version:
return None
results = SMCOGResults(json["record_id"])
for hit, parts in json["best_hits"].items():
results.best_hits[hit] = hmmscan_refinement.HMMResult(*parts)
if json.get("image_dir"):
results.relative_tree_path = json["image_dir"]
results.tree_images = json["tree_paths"]
return results
def from_json(json: Dict[str, Any], record: Record) -> "SMCOGResults":
if json.get("schema_version") != SMCOGResults.schema_version:
logging.debug("Schema version mismatch, discarding SMCOGs results")
return None
if record.id != json.get("record_id"):
logging.debug("Record ID mismatch, discarding SMCOGs results")
return None
results = SMCOGResults(json["record_id"])
for hit, parts in json["best_hits"].items():
results.best_hits[hit] = hmmscan_refinement.HMMResult(*parts)
if json.get("image_dir"):
results.relative_tree_path = json["image_dir"]
results.tree_images = json["tree_paths"]
return results
def test_json_conversion(self):
result = refinement.HMMResult("dummy_hit", 1, 5, 3e-10, 53.5)
json = result.to_json()
assert json == {
'bitscore': 53.5,
'evalue': 3e-10,
'hit_id': 'dummy_hit',
'query_end': 5,
'query_start': 1
}
regenerated = refinement.HMMResult.from_json(json)
assert regenerated.hit_id == "dummy_hit"
assert regenerated.query_start == 1
assert regenerated.query_end == 5
assert regenerated.evalue == 3e-10
assert regenerated.bitscore == 53.5
def test_incomplete_regulator(self):
results = refinement.gather_by_query(self.results)[self.gene_id]
results = sorted(list(results), key=lambda result: result.query_start)
assert len(results) == 5
regulator_id = "DUMMY:some_regulator_desc"
regulator_result = refinement.HMMResult(regulator_id, 1, 2, 1e-10, 1)
results.append(regulator_result)
new_lengths = dict(self.hmm_lengths)
new_lengths[regulator_id] = len(regulator_result) * 100 # always big
# set the thresholds to be unreachable
new = refinement._remove_incomplete(results,
new_lengths,
threshold=2.,
fallback=2.)
# ensure the tiny, but present, regulator is still in the list
assert len(new) == 1
assert new[0].hit_id == regulator_id
def test_str_conversion(self):
result = refinement.HMMResult("dummy_hit", 1, 5, 3e-10, 53.5)
assert str(
result
) == "HMMResult(dummy_hit, 1, 5, evalue=3e-10, bitscore=53.5)"
