def test_stop_without_start(self):
seq = "NNNTAGNNN"
for offset in [0, 6, 305]:
if offset == 0:
orfs = scan_orfs(seq, 1)
else:
orfs = scan_orfs(seq, 1, offset=offset)
assert len(orfs) == 1
orf = orfs[0]
assert isinstance(orf.start, BeforePosition)
assert isinstance(orf.end, ExactPosition)
assert orf.end == 6 + offset
def test_start_without_end(self):
seq = "NNNATGNNN"
for offset in [0, 6, 305]:
if offset == 0:
orfs = scan_orfs(seq, 1)
else:
orfs = scan_orfs(seq, 1, offset=offset)
assert len(orfs) == 1
orf = orfs[0]
assert isinstance(orf.start, ExactPosition)
assert isinstance(orf.end, AfterPosition)
assert orf.start == 3 + offset
def test_contained(self):
seq = "ATG" + "X" * 60 + "TAG"
for offset in [0, 6, 305]:
if offset == 0:
orfs = scan_orfs(seq, 1)
else:
orfs = scan_orfs(seq, 1, offset=offset)
assert len(orfs) == 1
orf = orfs[0]
assert isinstance(orf.start, ExactPosition)
assert isinstance(orf.end, ExactPosition)
assert orf.start == 0 + offset
assert orf.end == 66 + offset
def test_multiframe(self):
starts = [3, 25, 72, 77]
seq = ["X"] * 200
for start in starts:
seq[start:start + 3] = "ATG"
seq[start + 63:start + 66] = "TAA"
assert len(seq) == 200
result = scan_orfs("".join(seq), direction=1)
assert sorted([orf.start for orf in result]) == starts
assert len(result) == 4