def testSplit(self):
fq1 = StringIO.StringIO('''\
@foo/1 comment1
ACGTACGT
+
;;;;;;;;
''')
fq2 = StringIO.StringIO('''\
@foo/2 comment2
acgtacgt
+
AAAAAAAA
''')
out = StringIO.StringIO('')
ngsutils.fastq.merge.fastq_merge(
[FASTQ(fileobj=fq1), FASTQ(fileobj=fq2)],
split_slashes=True,
out=out,
quiet=True)
self.assertEqual(
out.getvalue(), '''\
@foo /1 comment1
ACGTACGT
+
;;;;;;;;
@foo /2 comment2
acgtacgt
+
AAAAAAAA
''')
def testSplitThree(self):
fname = os.path.join(os.path.dirname(__file__), 'test.fastq')
templ = os.path.join(os.path.dirname(__file__), 'test_templ')
ngsutils.fastq.split.fastq_split(fname,
templ,
3,
ignore_pairs=True,
quiet=True)
self.assertTrue(os.path.exists('%s.1.fastq' % templ))
self.assertTrue(os.path.exists('%s.2.fastq' % templ))
self.assertTrue(os.path.exists('%s.3.fastq' % templ))
fq1 = FASTQ('%s.1.fastq' % templ)
fq2 = FASTQ('%s.2.fastq' % templ)
fq3 = FASTQ('%s.3.fastq' % templ)
names1 = [x.fullname for x in fq1.fetch(quiet=True)]
self.assertEqual(names1, ['foo /1', 'bar /2'])
names2 = [x.fullname for x in fq2.fetch(quiet=True)]
self.assertEqual(names2, ['foo /2', 'baz /1'])
names3 = [x.fullname for x in fq3.fetch(quiet=True)]
self.assertEqual(names3, ['bar /1', 'baz /2'])
fq1.close()
fq2.close()
fq3.close()
os.unlink('%s.1.fastq' % templ)
os.unlink('%s.2.fastq' % templ)
os.unlink('%s.3.fastq' % templ)
def testSplitUnpaired(self):
fname = os.path.join(os.path.dirname(__file__), 'test.fastq')
templ = os.path.join(os.path.dirname(__file__), 'test_templ')
ngsutils.fastq.split.fastq_split(fname,
templ,
2,
ignore_pairs=True,
quiet=True)
self.assertTrue(os.path.exists('%s.1.fastq' % templ))
self.assertTrue(os.path.exists('%s.2.fastq' % templ))
fq1 = FASTQ('%s.1.fastq' % templ)
fq2 = FASTQ('%s.2.fastq' % templ)
names1 = [x.name for x in fq1.fetch(quiet=True)]
self.assertEqual(names1, ['foo', 'bar', 'baz'])
names2 = [x.name for x in fq2.fetch(quiet=True)]
self.assertEqual(names2, ['foo', 'bar', 'baz'])
fq1.close()
fq2.close()
os.unlink('%s.1.fastq' % templ)
os.unlink('%s.2.fastq' % templ)
def testMissingBoth(self):
fq1 = StringIO.StringIO('''\
@foo comment
ACGTACGT
+
;;;;;;;;
@bar
ACGTACGT
+
;;;;;;;;
@quux
ACGTACGT
+
;;;;;;;;
''')
fq2 = StringIO.StringIO('''\
@foo comment
ACGTACGT
+
;;;;;;;;
@baz
ACGTACGT
+
;;;;;;;;
@quux
ACGTACGT
+
;;;;;;;;
''')
out1 = StringIO.StringIO('')
out2 = StringIO.StringIO('')
fastq1 = FASTQ(fileobj=fq1)
fastq2 = FASTQ(fileobj=fq2)
ngsutils.fastq.properpairs.find_fastq_pairs(fastq1,
fastq2,
out1,
out2,
quiet=True)
self.assertEqual(out1.getvalue(), out2.getvalue())
self.assertEqual(
out1.getvalue(), '''\
@foo comment
ACGTACGT
+
;;;;;;;;
@quux
ACGTACGT
+
;;;;;;;;
''')
def fastq_unmerge(combined_fname, out_template, gz=False):
outs = []
if gz:
outs.append(gzip.open('%s.1.fastq.gz' % out_template, 'w'))
else:
outs.append(open('%s.1.fastq' % out_template, 'w'))
outidx = 1
last_read = None
fq = FASTQ(combined_fname)
for read in fq.fetch():
if last_read and last_read.name == read.name:
outidx += 1
if len(outs) < outidx:
if gz:
outs.append(
gzip.open('%s.%s.fastq.gz' % (out_template, outidx),
'w'))
else:
outs.append(
open('%s.%s.fastq' % (out_template, outidx), 'w'))
read.write(outs[outidx - 1])
else:
outidx = 1
read.write(outs[0])
last_read = read
fq.close()
for out in outs:
out.close()
def testFilterQualIllumina(self):
fq = StringIO.StringIO('''\
@foo
ACGTACGTACGTACGT
+
JJJJJJJJJEEEEJJJ
@bar
ACGTACGTA
+
JJJJJEEJJ
''')
out = StringIO.StringIO('')
chain = ngsutils.fastq.filter.FASTQReader(FASTQ(fileobj=fq),
verbose=False)
chain = ngsutils.fastq.filter.QualFilter(chain,
7,
5,
illumina=True,
verbose=False)
ngsutils.fastq.filter.fastq_filter(chain, out=out, quiet=True)
self.assertEqual(
out.getvalue(), '''\
@foo #qual
ACGTACGTACGT
+
JJJJJJJJJEEE
@bar
ACGTACGTA
+
JJJJJEEJJ
''')
def testTrimMinOffset(self):
fq = StringIO.StringIO('''\
@foo
aaaaaTTGCaaccggttccttggaa
+
;;;;;;;;;;;;;;;;;;;;;;;;;
@bar
TTGCtgtgatagctacgactaaaacc
+
;;;;;;;;;;;;;;;;;;;;;;;;;;
''')
out = StringIO.StringIO('')
ngsutils.fastq.trim.fastq_trim(FASTQ(fileobj=fq),
linker_5='TTGC',
min_len=20,
quiet=True,
out=out)
self.assertEqual(
out.getvalue(), '''\
@foo
aaaaaTTGCaaccggttccttggaa
+
;;;;;;;;;;;;;;;;;;;;;;;;;
@bar
tgtgatagctacgactaaaacc
+
;;;;;;;;;;;;;;;;;;;;;;
''')
def testTrimPctIdentity0_8(self):
fq = StringIO.StringIO('''\
@foo
TTGCCaaccggttccttagaa
+
;;;;;;;;;;;;;;;;;;;;;
@bar
TTGGCtgtgatagctacgactaaaacc
+
;;;;;;;;;;;;;;;;;;;;;;;;;;;
''')
out = StringIO.StringIO('')
ngsutils.fastq.trim.fastq_trim(FASTQ(fileobj=fq),
linker_5='TTGGC',
min_len=10,
pct_identity=0.75,
quiet=True,
out=out)
self.assertEqual(
out.getvalue(), '''\
@foo
aaccggttccttagaa
+
;;;;;;;;;;;;;;;;
@bar
tgtgatagctacgactaaaacc
+
;;;;;;;;;;;;;;;;;;;;;;
''')
def testTrimCS(self):
# Note: colorspace trimming should really only be applicable to the 3' end
fq = StringIO.StringIO('''\
@foo
T012301231231013101112231
+
;;;;;;;;;;;;;;;;;;;;;;;;
@bar
T012301231231013101112111
+
;;;;;;;;;;;;;;;;;;;;;;;;
''')
out = StringIO.StringIO('')
ngsutils.fastq.trim.fastq_trim(FASTQ(fileobj=fq),
linker_3='112231',
min_len=10,
quiet=True,
out=out)
self.assertEqual(
out.getvalue(), '''\
@foo
T012301231231013101
+
;;;;;;;;;;;;;;;;;;
@bar
T012301231231013101112111
+
;;;;;;;;;;;;;;;;;;;;;;;;
''')
def testFilterTrimRepeat(self):
'''properly trim repeats'''
fq = StringIO.StringIO('''\
@foo
ACGTACGTAAAAAAAAAA
+
;;;;;;;;;;;;;;;;;;
@bar
CCGGATTAGGCCCAAA
+
;;;;;;;;;;;;;;;;
''')
out = StringIO.StringIO('')
chain = ngsutils.fastq.filter.FASTQReader(FASTQ(fileobj=fq),
verbose=False)
chain = ngsutils.fastq.filter.TrimFilter(chain,
'AAAAAAAAAAAAAAAAAA',
1.0,
4,
verbose=False)
ngsutils.fastq.filter.fastq_filter(chain, out=out, quiet=True)
self.assertEqual(
out.getvalue(), '''\
@foo #trim
ACGTACGT
+
;;;;;;;;
@bar
CCGGATTAGGCCCAAA
+
;;;;;;;;;;;;;;;;
''')
def testFilterSuffixQual(self):
fq = StringIO.StringIO('''\
@foo
ACGTACGTACGTATTT
+
;;;;;;;;;;;;!!!!
@bar
ACGTACGTA
+
;;;;;;;;!
''')
out = StringIO.StringIO('')
chain = ngsutils.fastq.filter.FASTQReader(FASTQ(fileobj=fq),
verbose=False)
chain = ngsutils.fastq.filter.SuffixQualFilter(chain,
'!',
verbose=False)
ngsutils.fastq.filter.fastq_filter(chain, out=out, quiet=True)
self.assertEqual(
out.getvalue(), '''\
@foo #suff
ACGTACGTACGT
+
;;;;;;;;;;;;
@bar #suff
ACGTACGT
+
;;;;;;;;
''')
def fastq_tile(fname, outbase, length, offset, gz=False, quiet=False):
fastq = FASTQ(fname)
outs = []
fnames = []
for read in fastq.fetch(quiet=quiet):
out_idx = 0
pos = 0
while pos + length < len(read.seq):
if len(outs) <= out_idx:
fobj, tmp, fn = _open_file(outbase, out_idx, gz, quiet)
outs.append(fobj)
fnames.append((tmp, fn))
read.subseq(pos,
pos + length,
comment="#tile:%s,%s" % (pos, pos + length)).write(
outs[out_idx])
pos += offset
out_idx += 1
for out in outs:
out.close()
fastq.close()
for tmp, fname in fnames:
os.rename(tmp, fname)
def testFilterDiscard(self):
fq = StringIO.StringIO('''\
@foo
ACGTACGTACGTACGT
+
;;;;;;;;;;;;;;;;
@bar
ACGTACGT
+
;;;;;;;;
''')
discarded = []
def _discard(name):
discarded.append(name)
out = StringIO.StringIO('')
chain = ngsutils.fastq.filter.FASTQReader(FASTQ(fileobj=fq),
verbose=False)
chain = ngsutils.fastq.filter.SizeFilter(chain,
12,
verbose=False,
discard=_discard)
ngsutils.fastq.filter.fastq_filter(chain, out=out, quiet=True)
self.assertEqual(discarded, ['bar'])
def testFilterQual(self):
fq = StringIO.StringIO('''\
@foo comment
ACGTACGTACGTACGT
+
+++++++++&&&&+++
@bar
ACGTACGTA
+
+++++&&++
''')
out = StringIO.StringIO('')
chain = ngsutils.fastq.filter.FASTQReader(FASTQ(fileobj=fq),
verbose=False)
chain = ngsutils.fastq.filter.QualFilter(chain, 7, 5, verbose=False)
ngsutils.fastq.filter.fastq_filter(chain, out=out, quiet=True)
self.assertEqual(
out.getvalue(), '''\
@foo comment #qual
ACGTACGTACGT
+
+++++++++&&&
@bar
ACGTACGTA
+
+++++&&++
''')
def fastq_split(fname,
outbase,
chunks,
ignore_pairs=False,
gz=False,
count_fname=None,
quiet=False):
fastq = FASTQ(fname)
if ignore_pairs:
is_paired = False
else:
is_paired = fastq.is_paired
outs = []
fnames = []
for i in xrange(chunks):
if gz:
fn = '%s.%s.fastq.gz' % (outbase, i + 1)
tmp = os.path.join(os.path.dirname(fn),
'.tmp.%s' % os.path.basename(fn))
fnames.append((tmp, fn))
if not quiet:
sys.stderr.write('Output file: %s\n' % fn)
outs.append(gzip.open(tmp, 'w'))
else:
fn = '%s.%s.fastq' % (outbase, i + 1)
tmp = os.path.join(os.path.dirname(fn),
'.tmp.%s' % os.path.basename(fn))
fnames.append((tmp, fn))
if not quiet:
sys.stderr.write('Output file: %s\n' % fn)
outs.append(open(tmp, 'w'))
i = chunks
last_name = None
for read in fastq.fetch(quiet=quiet):
if not is_paired:
i += 1
elif read.name != last_name:
i += 1
if i >= len(outs):
i = 0
last_name = read.name
read.write(outs[i])
for out in outs:
out.close()
fastq.close()
for tmp, fname in fnames:
os.rename(tmp, fname)
def testFQRead(self):
fq = StringIO.StringIO('''\
@foo
ACGTacgtACGT
+
CDEFGHIJKLMN
''')
out = StringIO.StringIO('')
ngsutils.fastq.convertqual.fastq_convertqual(FASTQ(fileobj=fq),
out=out,
quiet=True)
out.seek(0)
fqout = FASTQ(fileobj=out)
read = fqout.fetch().next()
self.assertEqual(read.name, 'foo')
self.assertEqual(read.seq, 'ACGTacgtACGT')
self.assertEqual(read.qual, "$%&'()*+,-./")
def testMerge(self):
fq1 = StringIO.StringIO('''\
@foo
ACGTACGT
+
;;;;;;;;
@bar comment
ACGTACGT
+
;;;;;;;;
''')
fq2 = StringIO.StringIO('''\
@foo
acgtacgt
+
AAAAAAAA
@bar comment
acgtacgt
+
AAAAAAAA
''')
out = StringIO.StringIO('')
ngsutils.fastq.merge.fastq_merge(
[FASTQ(fileobj=fq1), FASTQ(fileobj=fq2)], out=out, quiet=True)
self.assertEqual(
out.getvalue(), '''\
@foo
ACGTACGT
+
;;;;;;;;
@foo
acgtacgt
+
AAAAAAAA
@bar comment
ACGTACGT
+
;;;;;;;;
@bar comment
acgtacgt
+
AAAAAAAA
''')
def test_splitFastq(self):
path = os.path.dirname(__file__)
ngsutils.fastq.barcode_split.fastx_barcode_split(FASTQ(os.path.join(path, 'test_barcodes.fastq')), os.path.join(path, 'out.%s.fastq'), barcodes2, allow_revcomp=True)
self.assert_fastq_contains(os.path.join(path, 'out.%s.fastq'), {
'missing': ('quux', '', ''),
'tag1': ('foo foo-rc', 'atcgatcgatcgatcg atcgatcgatcgatcg', 'AAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAA'),
'tag2': ('bar', 'gctagctagctagcta', 'AAAAAAAAAAAAAAAA'),
'tag3': ('baz', 'acgtacgtacgtacgt', 'AAAAAAAAAAAAAAAA')
})
self._unlink_fastx(os.path.join(path, 'out.%s.fastq'), 'missing tag1 tag2 tag3'.split())
def testAssert(self):
fq1 = StringIO.StringIO('''\
@foo
ACGTACGT
+
;;;;;;;;
''')
fq2 = StringIO.StringIO('''\
@bar
acgtacgt
+
AAAAAAAA
''')
out = StringIO.StringIO('')
self.assertRaises(ValueError, ngsutils.fastq.merge.fastq_merge, *[
[FASTQ(fileobj=fq1), FASTQ(fileobj=fq2)],
], **{
'out': out,
'quiet': True
})
def testFQRead(self):
fq = StringIO.StringIO('''\
@foo
T0000111122223333....
+
....................
''')
out = StringIO.StringIO('')
ngsutils.fastq.csencode.fastq_csencode(FASTQ(fileobj=fq), out=out, quiet=True)
self.assertEqual('''@foo
AAACCCCGGGGTTTTNNNN
+
...................
''', out.getvalue())
def testFilterPaired(self):
fq = StringIO.StringIO('''\
@foo
ACGTACGTACGTACGT
+
;;;;;;;;;;;;;;;;
@foo
ACGTACGTACGTACGT
+
;;;;;;;;;;;;;;;;
@bar
ACGTACGT
+
;;;;;;;;
@baz
ACGTACGT
+
;;;;;;;;
@baz
ACGTACGT
+
;;;;;;;;
''')
out = StringIO.StringIO('')
chain = ngsutils.fastq.filter.FASTQReader(FASTQ(fileobj=fq),
verbose=False)
chain = ngsutils.fastq.filter.PairedFilter(chain, verbose=False)
ngsutils.fastq.filter.fastq_filter(chain, out=out, quiet=True)
self.assertEqual(
out.getvalue(), '''\
@foo
ACGTACGTACGTACGT
+
;;;;;;;;;;;;;;;;
@foo
ACGTACGTACGTACGT
+
;;;;;;;;;;;;;;;;
@baz
ACGTACGT
+
;;;;;;;;
@baz
ACGTACGT
+
;;;;;;;;
''')
def testNames(self):
fq = StringIO.StringIO('''\
@foo
ACGTACGT
+
;;;;;;;;
@bar comment
ACGTACGT
+
;;;;;;;;
''')
out = StringIO.StringIO('')
ngsutils.fastq.names.export_names(FASTQ(fileobj=fq),
out=out,
quiet=True)
self.assertEqual(out.getvalue(), 'foo\nbar\n')
def testFilterTrimCS(self):
fq = StringIO.StringIO('''\
@foo
T0123012301231122
+
;;;;;;;;;;;;;;;;
@bar
T012301231122
+
;;;;;;;;;;;;
@baz
T012301231102
+
;;;;;;;;;;;;
@quux
T1122Atcgtagt
+
;;;;;;;;;;;;
''')
out = StringIO.StringIO('')
chain = ngsutils.fastq.filter.FASTQReader(FASTQ(fileobj=fq),
verbose=False)
chain = ngsutils.fastq.filter.TrimFilter(chain,
'1122',
0.8,
3,
verbose=False)
ngsutils.fastq.filter.fastq_filter(chain, out=out, quiet=True)
self.assertEqual(
out.getvalue(), '''\
@foo #trim
T012301230123
+
;;;;;;;;;;;;
@bar #trim
T01230123
+
;;;;;;;;
@baz
T012301231102
+
;;;;;;;;;;;;
''')
def testFilterTrim(self):
fq = StringIO.StringIO('''\
@foo
ACGTACGTACGTATTT
+
;;;;;;;;;;;;;;;;
@bar
ACGTatttACGT
+
;;;;;;;;;;;;
@baz
ACGTACGTATTC
+
;;;;;;;;;;;;
@quux
ATTTAtcgtagt
+
;;;;;;;;;;;;
''')
out = StringIO.StringIO('')
chain = ngsutils.fastq.filter.FASTQReader(FASTQ(fileobj=fq),
verbose=False)
chain = ngsutils.fastq.filter.TrimFilter(chain,
'ATTT',
1.0,
3,
verbose=False)
ngsutils.fastq.filter.fastq_filter(chain, out=out, quiet=True)
self.assertEqual(
out.getvalue(), '''\
@foo #trim
ACGTACGTACGT
+
;;;;;;;;;;;;
@bar #trim
ACGT
+
;;;;
@baz
ACGTACGTATTC
+
;;;;;;;;;;;;
''')
def assert_fastq_contains(self, base, args):
for tag in args:
valid = args[tag][0].split()
seq_qual = {}
if args[tag][1]:
for n, s, q in zip(valid, args[tag][1].split(), args[tag][2].split()):
seq_qual[n] = (s, q)
fq = FASTQ(base % tag)
count = 0
for read in fq.fetch():
if read.name in valid:
count += 1
if seq_qual:
self.assertEqual(seq_qual[read.name], (read.seq, read.qual))
else:
self.assertEqual('extra read in %s' % tag, read.name)
self.assertEqual(count, len(valid))
def testFilterTrim2(self):
'''Test to make sure that we don't trim out the middle of a read'''
fq = StringIO.StringIO('''\
@foo
ACGTACGTACGTATTATT
+
;;;;;;;;;;;;;;;;;;
@bar
CCGGATTAGGCCGGCC
+
;;;;;;;;;;;;;;;;
@baz
CCGGATTATTATTATT
+
;;;;;;;;;;;;;;;;
''')
out = StringIO.StringIO('')
chain = ngsutils.fastq.filter.FASTQReader(FASTQ(fileobj=fq),
verbose=False)
chain = ngsutils.fastq.filter.TrimFilter(chain,
'ATTATT',
1.0,
4,
verbose=False)
ngsutils.fastq.filter.fastq_filter(chain, out=out, quiet=True)
self.assertEqual(
out.getvalue(), '''\
@foo #trim
ACGTACGTACGT
+
;;;;;;;;;;;;
@bar
CCGGATTAGGCCGGCC
+
;;;;;;;;;;;;;;;;
@baz #trim
CCGGATTATT
+
;;;;;;;;;;
''')
def testSimple(self):
fq = StringIO.StringIO('''\
@foo
ACGTACGT
+
;;;;;;;;
@bar
ACGTACGT
+
;;;;;;;;
@baz
ACGTACGTAC
+
;;;;;;;;;;
@quux
ACGTACGTAC
+
;;;;;;;;;;
''')
# two reads length 8, two reads length 10
stats = ngsutils.fastq.stats.fastq_stats(FASTQ(fileobj=fq), quiet=True)
self.assertEqual(stats.total_reads, 4)
self.assertEqual(stats.lengths, [0, 0, 0, 0, 0, 0, 0, 0, 2, 0, 2])
self.assertEqual(stats.totals, [0, 4, 4, 4, 4, 4, 4, 4, 4, 2, 2])
self.assertEqual(
stats.qualities,
[0, 104, 104, 104, 104, 104, 104, 104, 104, 52, 52]) # accumulator
for q, t in zip(stats.qualities, stats.totals)[1:]:
self.assertEqual(q / t, 26)
self.assertEqual(stats.length_stats.mean, 9)
self.assertEqual(stats.length_stats.min_val, 8)
self.assertEqual(stats.length_stats.max_val, 10)
self.assertEqual(len(stats.quality_stats), 11)
for qvstats in stats.quality_stats:
if not qvstats:
continue
self.assertEqual(qvstats.mean, 26)
def testTruncateCS(self):
fq = StringIO.StringIO('''\
@foo
T01230123
+
;;;;;;;;
@bar comment
T01230123
+
;;;;;;;;
''')
out = StringIO.StringIO('')
ngsutils.fastq.truncate.fastq_truncate(FASTQ(fileobj=fq), 4, quiet=True, out=out)
self.assertEqual(out.getvalue(), '''\
@foo
T0123
+
;;;;
@bar comment
T0123
+
;;;;
''')
def testTrim53(self):
fq = StringIO.StringIO('''\
@foo
TTGCaaccggttccttggaaACGT
+
;;;;;;;;;;;;;;;;;;;;;;;;
@bar - note offset
aaTTGCtgtgatagctacgactACGTa
+
;;;;;;;;;;;;;;;;;;;;;;;;;;;
@baz
TAGCtgtagatgatagatagaAGCT
+
;;;;;;;;;;;;;;;;;;;;;;;;;
''')
out = StringIO.StringIO('')
ngsutils.fastq.trim.fastq_trim(FASTQ(fileobj=fq),
linker_5='TTGC',
linker_3='ACGT',
min_len=10,
quiet=True,
out=out)
self.assertEqual(
out.getvalue(), '''\
@foo
aaccggttccttggaa
+
;;;;;;;;;;;;;;;;
@bar - note offset
tgtgatagctacgact
+
;;;;;;;;;;;;;;;;
@baz
TAGCtgtagatgatagatagaAGCT
+
;;;;;;;;;;;;;;;;;;;;;;;;;
''')
def testFilterWildcard(self):
fq = StringIO.StringIO('''\
@foo
ACGTACNNACGTATTT
+
;;;;;;;;;;;;;;;;
@bar
ACGTACGT
+
;;;;;;;;
@baz
ACGTACGNACTG
+
;;;;;;;;;;;;
@quux
ACG..CGAACTG
+
;;;;;;;;;;;;
''')
out = StringIO.StringIO('')
chain = ngsutils.fastq.filter.FASTQReader(FASTQ(fileobj=fq),
verbose=False)
chain = ngsutils.fastq.filter.WildcardFilter(chain, 1, verbose=False)
ngsutils.fastq.filter.fastq_filter(chain, out=out, quiet=True)
self.assertEqual(
out.getvalue(), '''\
@bar
ACGTACGT
+
;;;;;;;;
@baz
ACGTACGNACTG
+
;;;;;;;;;;;;
''')