def test_get_sequence(self):
"""get the sequence"""
exp_gb1_start = "CCATGATTCGACCATTTTCAGAGAG"
exp_gb1_end = "CAACGGTCAGGCCAG"
exp_gb1_len = 452612
obs = get_sequence(self.gb1)
obs_gb1_start = obs[:25]
obs_gb1_end = obs[-15:]
obs_gb1_len = len(obs)
self.assertEqual(obs_gb1_start, exp_gb1_start)
self.assertEqual(obs_gb1_end, exp_gb1_end)
self.assertEqual(obs_gb1_len, exp_gb1_len)
# accession is str including version
try:
accession = get_accession(next_record)
except:
failure_count += 1
continue
if accession in observed_records:
continue
if accession in seen:
continue
# accession added at the end of this while loop
# sequence is just a str of sequence
try:
sequence = get_sequence(next_record)
except NoSequenceError:
# this isn't a failure, so no point in continuing but record
# the accession so it isn't hit again
write_obs_record(obs_records, accession)
continue
except:
failure_count += 1
continue
# verify the sequence is DNA. NCBI silently corrupts records
# every once and a while leading to crap in the sequence.
# Thanks NCBI.
seq_chars = set(sequence)
if not seq_chars.issubset(alpha):
logline = log_f("Corrupt sequence, accession: %s" % (accession))
