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 test_only_sequences(self):
"""RdbParser should return empty list form file w/o lables"""
# should fail if strict (the default)
self.assertRaises(RecordError, list,
RdbParser(self.nolabels, strict=True))
# if not strict, should skip the records
self.assertEqual(list(RdbParser(self.nolabels, strict=False)), [])
def test_single_constructor(self):
"""RdbParser should use constructors if supplied"""
to_dna = lambda x, info: DnaSequence(str(x).replace("U", "T"), info=info)
f = list(RdbParser(self.oneseq, to_dna))
self.assertEqual(len(f), 1)
a = f[0]
self.assertEqual(a, "AGTCATCTAGATHCATHC")
self.assertEqual(
a.info, Info({"Species": "H.Sapiens", "OriginalSeq": "AGUCAUCUAGAUHCAUHC"})
)
def alternativeConstr(header_lines):
info = Info()
for line in header_lines:
all = line.strip().split(":", 1)
# strip out empty lines, lines without name, lines without
# colon
if not all[0] or len(all) != 2:
continue
name = all[0].upper()
value = all[1].strip().upper()
info[name] = value
return info
f = list(RdbParser(self.oneseq, to_dna, alternativeConstr))
self.assertEqual(len(f), 1)
a = f[0]
self.assertEqual(a, "AGTCATCTAGATHCATHC")
exp_info = Info(
{"OriginalSeq": "AGUCAUCUAGAUHCAUHC", "Refs": {}, "SEQ": "H.SAPIENS"}
)
self.assertEqual(
a.info,
Info({"OriginalSeq": "AGUCAUCUAGAUHCAUHC", "Refs": {}, "SEQ": "H.SAPIENS"}),
)
def test_full(self):
"""RdbParser: full data, valid and invalid"""
# when only good record, should work independent of strict
r1 = RnaSequence(
"-??GG-UGAA--CGCU---ACGU-N???---",
info=Info({
"Species": "unidentified Thermus OPB AF027020",
"Refs": {
"rRNA": ["AF027020"]
},
"OriginalSeq": "-o[oGG-U{G}AA--C^GC]U---ACGU-Nooo---",
}),
)
r2 = RnaSequence(
"---CGAUCG--UAUACG-N???-",
info=Info({
"Species": "Thermus silvanus X84211",
"Refs": {
"rRNA": ["X84211"]
},
"OriginalSeq": "---CGAU[C(G){--UA}U]ACG-Nooo-",
}),
)
obs = list(RdbParser(RDB_LINES_ONLY_GOOD.split("\n"), strict=True))
self.assertEqual(len(obs), 2)
self.assertEqual(obs[0], r1)
self.assertEqual(str(obs[0]), str(r1))
self.assertEqual(obs[0].info, r1.info)
self.assertEqual(obs[1], r2)
self.assertEqual(str(obs[1]), str(r2))
self.assertEqual(obs[1].info, r2.info)
obs = list(RdbParser(RDB_LINES_ONLY_GOOD.split("\n"), strict=False))
self.assertEqual(len(obs), 2)
self.assertEqual(obs[0], r1)
self.assertEqual(str(obs[0]), str(r1))
self.assertEqual(obs[0].info, r1.info)
# when strict, should raise error on invalid record
f = RdbParser(RDB_LINES_GOOD_BAD.split("\n"), strict=True)
self.assertRaises(RecordError, list, f)
# when not strict, malicious record is skipped
obs = list(RdbParser(RDB_LINES_GOOD_BAD.split("\n"), strict=False))
self.assertEqual(len(obs), 2)
self.assertEqual(obs[0], r1)
self.assertEqual(str(obs[0]), str(r1))
self.assertEqual(obs[0].info, r1.info)
self.assertEqual(obs[1], r2)
self.assertEqual(str(obs[1]), str(r2))
self.assertEqual(obs[1].info, r2.info)
def test_multiple_constructor_bad(self):
"""RdbParser should complain or skip bad records w/ constructor"""
def dnastrict(x, **kwargs):
try:
return DnaSequence(x, **kwargs)
except Exception:
raise RecordError("Could not convert sequence")
self.assertRaises(RecordError, list, RdbParser(self.oneX, dnastrict))
f = list(RdbParser(self.oneX, dnastrict, strict=False))
self.assertEqual(len(f), 2)
a, b = f
self.assertEqual(a, "ACT")
self.assertEqual(a.info, Info({"Species": "mit", "OriginalSeq": "ACT"}))
self.assertEqual(b, "AAA")
self.assertEqual(b.info, Info({"Species": "pla", "OriginalSeq": "AAA"}))
def test_empty(self):
"""RdbParser should return empty list from 'file' w/o labels"""
self.assertEqual(list(RdbParser(self.empty)), [])
self.assertEqual(list(RdbParser(self.nolabels, strict=False)), [])
self.assertRaises(RecordError, list, RdbParser(self.nolabels))
