def test_single(self):
"""RdbParser should read single record as (header,seq) tuple"""
res = list(RdbParser(self.oneseq))
self.assertEqual(len(res), 1)
first = res[0]
self.assertEqual(first, Sequence('AGUCAUCUAGAUHCAUHC'))
self.assertEqual(first.Info, Info({'Species':'H.Sapiens',\
'OriginalSeq':'AGUCAUCUAGAUHCAUHC'}))
res = list(RdbParser(self.multiline))
self.assertEqual(len(res), 1)
first = res[0]
self.assertEqual(first, Sequence('AGUCAUUAGAUHCAUHC'))
self.assertEqual(first.Info, Info({'Species':'H.Sapiens',\
'OriginalSeq':'AGUCAUUAGAUHCAUHC'}))
def toSeq(self, Bases=True, truncate=True):
"""Translates flowgram to sequence and returns sequence object
if Bases is True then a sequence object will be made using
self.Bases instead of translating the flowgram
truncate: if True strip off lowercase chars (low quality bases)
"""
if Bases and hasattr(self, "Bases"):
seq = self.Bases
else:
seq = []
if self.floworder is None:
raise ValueError, "must have self.floworder set"
key = FakeRandom(self.floworder, True)
flows_since_last = 0
for n in self.flowgram:
signal = int(round(n))
seq.extend([key()] * signal)
if (signal > 0):
flows_since_last = 0
else:
flows_since_last += 1
if (flows_since_last == 4):
seq.extend('N')
flows_since_last = 0
seq = ''.join(seq)
#cache the result for next time
self.Bases = seq
if (truncate):
seq = str(seq)
seq = seq.rstrip("acgtn")
seq = seq.lstrip("actgn")
return Sequence(seq, Name=self.Name)
def test_init_other_seq(self):
"""Sequence init with other seq should preserve name and info."""
r = self.RNA('UCAGG', Name='x', Info={'z': 3})
s = Sequence(r)
self.assertEqual(s._seq, 'UCAGG')
self.assertEqual(s.Name, 'x')
self.assertEqual(s.Info.z, 3)
def test_gapped_to_ungapped_general_seq(self):
"""gapped_to_ungapped: when input is Sequence obj, treat as string
"""
s = Sequence(self.gapped)
p = self.simple_g
obs_seq, obs_pairs = gapped_to_ungapped(s, p)
self.assertEqual(obs_seq, self.ungapped)
self.assertEqualItems(obs_pairs, self.simple)
#assert not isinstance(obs_seq, Sequence)
#assert isinstance(obs_seq, str)
assert isinstance(obs_seq, Sequence)
assert isinstance(obs_pairs, Pairs)