def assert_fasta_contains(self, base, args):
for tag in args:
valid = args[tag].split()
fa = FASTA(base % tag)
count = 0
for read in fa.fetch():
if read.name in valid:
count += 1
else:
self.assertEqual('extra read in %s' % tag, read.name)
self.assertEqual(count, len(valid))
def testError2(self):
fa = StringIO.StringIO('''\
>foo_F3
ACTGACTG
''')
qa = StringIO.StringIO('''\
>foo_F3
10 10 10 20 20 20 20 A
''')
fasta = FASTA(fileobj=fa)
qual = FASTA(fileobj=qa)
out = StringIO.StringIO('')
self.assertRaises(ValueError, ngsutils.fastq.fromfasta.merge_files, *(fasta, qual), **{'out': out})
def test_splitFastaOffset(self):
path = os.path.dirname(__file__)
ngsutils.fastq.barcode_split.fastx_barcode_split(FASTA(os.path.join(path, 'test_barcodes.fasta')), os.path.join(path, 'out.%s.fasta'), barcodes2, allow_revcomp=True, pos=1)
self.assert_fasta_contains(os.path.join(path, 'out.%s.fasta'), {
'missing': 'foo1 bar1 baz1 foo1-rc',
'tag1': 'foo foo-rc foo2 foo2-rc',
'tag2': 'bar bar2',
'tag3': 'baz baz2'
})
self._unlink_fastx(os.path.join(path, 'out.%s.fasta'), 'missing tag1 tag2 tag3'.split())
def test_splitFasta(self):
path = os.path.dirname(__file__)
ngsutils.fastq.barcode_split.fastx_barcode_split(FASTA(os.path.join(path, 'test_barcodes.fasta')), os.path.join(path, 'out.%s.fasta'), barcodes2)
self.assert_fasta_contains(os.path.join(path, 'out.%s.fasta'), {
'missing': 'foo-rc foo1 bar1 baz1 foo2 bar2 baz2 foo1-rc foo2-rc',
'tag1': 'foo',
'tag2': 'bar',
'tag3': 'baz'
})
self._unlink_fastx(os.path.join(path, 'out.%s.fasta'), 'missing tag1 tag2 tag3'.split())
def testQualFile(self):
fa = StringIO.StringIO('''\
>foo_F3
ACTGACTG
''')
qa = StringIO.StringIO('''\
>foo_F3
10 10 10 10 0 93 -5 99
''')
fasta = FASTA(fileobj=fa)
qual = FASTA(fileobj=qa, qual=True)
out = StringIO.StringIO('')
ngsutils.fastq.fromfasta.merge_files(fasta, qual, out=out)
self.assertEqual(out.getvalue(), '''\
@foo
ACTGACTG
+
++++!~!~
''')
def testColor(self):
fa = StringIO.StringIO('''\
>foo_F3
T01230123
''')
qa = StringIO.StringIO('''\
>foo_F3
10 10 10 10 20 20 20 20
''')
fasta = FASTA(fileobj=fa)
qual = FASTA(fileobj=qa)
out = StringIO.StringIO('')
ngsutils.fastq.fromfasta.merge_files(fasta, qual, out=out)
self.assertEqual(out.getvalue(), '''\
@foo
T01230123
+
++++5555
''')
def testSuffix(self):
fa = StringIO.StringIO('''\
>foo_F3
ACTGACTG
''')
fasta = FASTA(fileobj=fa)
out = StringIO.StringIO('')
ngsutils.fastq.fromfasta.merge_files(fasta, None, common_qual=';', suffix='_bar', out=out)
self.assertEqual(out.getvalue(), '''\
@foo_bar
ACTGACTG
+
;;;;;;;;
''')
def testConstantColor(self):
fa = StringIO.StringIO('''\
>foo_F3
T01230123
''')
fasta = FASTA(fileobj=fa)
out = StringIO.StringIO('')
ngsutils.fastq.fromfasta.merge_files(fasta, None, common_qual=';', out=out)
self.assertEqual(out.getvalue(), '''\
@foo
T01230123
+
;;;;;;;;
''')
if first:
first = False
continue
cols = line.strip().split('\t')
if len(cols) < 4:
cols.append('y')
barcodes[cols[0]] = (cols[1], cols[2],
True if cols[3] in 'TtyY' else False)
elif not fname and os.path.exists(arg):
fname = arg
elif not templ:
templ = arg
if not barcodes or not fname or not templ:
usage()
if is_fasta_file(fname):
outtempl = '%s_%%s.fasta' % templ
reader = FASTA(fname)
else:
outtempl = '%s_%%s.fastq' % templ
reader = FASTQ(fname)
if stats:
stats_fname = '%s.stats.txt' % templ
else:
stats_fname = None
fastx_barcode_split(reader, outtempl, barcodes, edit, pos, allow_revcomp,
gz, stats_fname)
last = arg
elif arg == '-h':
usage()
elif not fasta and os.path.exists(arg):
fasta = arg
elif not qual and os.path.exists(arg):
qual = arg
else:
sys.stderr.write("ERROR: Unknown option: %s\n" % arg)
usage()
if not fasta:
usage()
if not qual and not common_qual:
sys.stderr.write(
'You must specify a quality input file or common qual value')
usage()
sys.stderr.write('Merging %s and %s\n' %
(os.path.basename(fasta),
os.path.basename(qual) if qual else common_qual))
f = FASTA(fasta)
q = FASTA(qual, qual=True) if qual else None
merge_files(f, q, tag, common_qual)
f.close()
if q:
q.close()
sys.exit(-1)
last = None
elif arg in ['-tag', '-q']:
last = arg
elif arg == '-h':
usage()
elif not fasta and os.path.exists(arg):
fasta = arg
elif not qual and os.path.exists(arg):
qual = arg
else:
sys.stderr.write("ERROR: Unknown option: %s\n" % arg)
usage()
if not fasta:
usage()
if not qual and not common_qual:
sys.stderr.write('You must specify a quality input file or common qual value')
usage()
sys.stderr.write('Merging %s and %s\n' % (os.path.basename(fasta), os.path.basename(qual) if qual else common_qual))
f = FASTA(fasta)
q = FASTA(qual, qual=True) if qual else None
merge_files(f, q, tag, common_qual)
f.close()
if q:
q.close()
