def test_validate_minabund():
filelist = kevlar.tests.data_glob('collect.beta.?.txt')
readset = ReadSet(19, 5e3)
for record in kevlar.seqio.afxstream(filelist):
readset.add(record)
readset.validate()
assert readset.valid == (4, 32)
readset = ReadSet(19, 5e3)
for record in kevlar.seqio.afxstream(filelist):
readset.add(record)
readset.validate(casemin=9)
assert readset.valid == (0, 0)
def test_ctrl3_refr_contam(bogusrefrcontam):
augfastq = kevlar.tests.data_file('trio1/novel_3_1,2.txt')
readset = ReadSet(13, 1e7, mask=bogusrefrcontam)
for record in parse_augmented_fastx(kevlar.open(augfastq, 'r')):
readset.add(record)
readset.validate()
assert readset.valid == (13, 171)
def test_ctrl3():
augfastq = kevlar.tests.data_file('trio1/novel_3_1,2.txt')
readset = ReadSet(13, 1e7)
for record in parse_augmented_fastx(kevlar.open(augfastq, 'r')):
readset.add(record)
readset.validate(casemin=6)
assert readset.valid == (424, 5782)
def test_load_readset():
filelist = kevlar.tests.data_glob('collect.beta.?.txt')
readset = ReadSet(19, 1e3)
for record in kevlar.seqio.afxstream(filelist):
readset.add(record)
assert len(readset) == 8
assert readset
kmers = [
'AGGGGCGTGACTTAATAAG', 'GGGCGTGACTTAATAAGGT',
'TAGGGGCGTGACTTAATAA', 'GGGGCGTGACTTAATAAGG',
]
for kmer in kmers:
assert readset._counts.get(kmer) == 8
def test_validate_with_mask():
kmer = 'AGGGGCGTGACTTAATAAG'
mask = khmer.Nodetable(19, 1e3, 2)
mask.add(kmer)
filelist = kevlar.tests.data_glob('collect.beta.?.txt')
readset = ReadSet(19, 5e3)
for record in kevlar.seqio.afxstream(filelist):
readset.add(record)
readset.validate(mask=mask)
assert readset.valid == (3, 24)
for record in readset:
for ikmer in record.ikmers:
assert ikmer.sequence != kmer
assert kevlar.revcom(ikmer.sequence) != kmer
def test_validate():
filelist = kevlar.tests.data_glob('collect.alpha.txt')
readset = ReadSet(19, 5e3)
for record in kevlar.seqio.afxstream(filelist):
readset.add(record)
readset.validate()
assert readset.valid == (4, 32)
assert len(readset) == 9
assert readset.discarded == 1
badkmers = ['CAGGCCAGGGATCGCCGTG']
goodkmers = [
'AGGGGCGTGACTTAATAAG', 'GGGCGTGACTTAATAAGGT',
'TAGGGGCGTGACTTAATAA', 'GGGGCGTGACTTAATAAGG',
]
for record in readset:
for kmer in record.ikmers:
assert kmer.sequence not in badkmers and \
kevlar.revcom(kmer.sequence) not in badkmers
assert kmer.sequence in goodkmers or \
kevlar.revcom(kmer.sequence) in goodkmers
def ctrl3():
augfastq = kevlar.tests.data_file('trio1/novel_3_1,2.txt')
readset = ReadSet(13, 1e7)
for record in kevlar.parse_augmented_fastx(kevlar.open(augfastq, 'r')):
readset.add(record)
return readset