def trimPrimers(primer, verbose):
"""
@param primer: A BioPython C{Bio.Seq} primer sequence.
@param verbose: A C{bool}, if C{True} output additional information about
how often and where primers were found.
"""
reads = []
absentCount = forwardCount = reverseCount = count = 0
for seqRecord in SeqIO.parse(sys.stdin, 'fasta'):
count += 1
start, end = findPrimerBidiLimits(primer, seqRecord.seq)
if start == 0:
if end == len(seqRecord):
absentCount += 1
else:
reverseCount += 1
else:
forwardCount += 1
if end != len(seqRecord):
reverseCount += 1
reads.append(seqRecord[start:end])
if verbose:
print(('Read %d sequences. Found forward: %d, '
'Found reversed: %d, Absent: %d') %
(count, forwardCount, reverseCount, absentCount),
file=sys.stderr)
SeqIO.write(reads, sys.stdout, 'fasta')
def testFoundMultiple(self):
"""
If a primer is found multiple times, the correct value
must be returned.
"""
seq = Seq('ACGTACGT', IUPAC.unambiguous_dna)
self.assertEqual((7, 8), findPrimerBidiLimits('ACG', seq))
def trimPrimers(primer, verbose):
"""
@param primer: A BioPython C{Bio.Seq} primer sequence.
@param verbose: A C{bool}, if C{True} output additional information about
how often and where primers were found.
"""
reads = []
absentCount = forwardCount = reverseCount = count = 0
for seqRecord in SeqIO.parse(sys.stdin, 'fasta'):
count += 1
start, end = findPrimerBidiLimits(primer, seqRecord.seq)
if start == 0:
if end == len(seqRecord):
absentCount += 1
else:
reverseCount += 1
else:
forwardCount += 1
if end != len(seqRecord):
reverseCount += 1
reads.append(seqRecord[start:end])
if verbose:
print((
'Read %d sequences. Found forward: %d, '
'Found reversed: %d, Absent: %d') % (
count, forwardCount, reverseCount, absentCount), file=sys.stderr)
SeqIO.write(reads, sys.stdout, 'fasta')
def testOverlappingBackwards(self):
"""
If a primer is present twice backwards but is overlapping, only
the first instance should be returned.
"""
seq = Seq('GTTT', IUPAC.unambiguous_dna)
self.assertEqual((0, 1), findPrimerBidiLimits('AA', seq))
def testNotFound(self):
"""
If a primer is not found, the returned offsets must include
the whole sequence.
"""
seq = Seq('ACGT', IUPAC.unambiguous_dna)
self.assertEqual((0, 4), findPrimerBidiLimits('BLAH', seq))
def testOverlappingBackwards(self):
"""
If a primer is present twice backwards but is overlapping, only
the first instance should be returned.
"""
seq = Seq('GTTT', IUPAC.unambiguous_dna)
self.assertEqual((0, 1), findPrimerBidiLimits('AA', seq))
def testFoundMultiple(self):
"""
If a primer is found multiple times, the correct value
must be returned.
"""
seq = Seq('ACGTACGT', IUPAC.unambiguous_dna)
self.assertEqual((7, 8), findPrimerBidiLimits('ACG', seq))
def testNotFound(self):
"""
If a primer is not found, the returned offsets must include
the whole sequence.
"""
seq = Seq('ACGT', IUPAC.unambiguous_dna)
self.assertEqual((0, 4), findPrimerBidiLimits('BLAH', seq))
def testOverlappingForwards(self):
"""
If a primer is present twice forwards but is overlapping, only
the first instance should be returned.
"""
seq = Seq('GAAA')
self.assertEqual((3, 4), findPrimerBidiLimits('AA', seq))
def testFoundEndStart(self):
"""
If a primer is found in both directions in a sequence (end of
the forward sequence, start of the reverse complement), the
correct value must be returned.
"""
seq = Seq('ACGT', IUPAC.unambiguous_dna)
self.assertEqual((4, 4), findPrimerBidiLimits('GT', seq))
def testFoundEndStart(self):
"""
If a primer is found in both directions in a sequence (end of
the forward sequence, start of the reverse complement), the
correct value must be returned.
"""
seq = Seq('ACGT', IUPAC.unambiguous_dna)
self.assertEqual((4, 4), findPrimerBidiLimits('GT', seq))
def testFoundStartEnd(self):
"""
If a primer is found in both directions in a sequence (start of
the forward sequence, end of the reverse complement), the
correct value must be returned.
"""
seq = Seq('ACGT')
self.assertEqual((2, 2), findPrimerBidiLimits('AC', seq))
def testLonger(self):
"""
Test a longer sequence.
"""
seq = Seq('AAAAAAAAAA'
'GGGGGGGGGG'
'AAAAAAAAAA'
'AAAAAAAAAA', IUPAC.unambiguous_dna)
self.assertEqual((20, 40), findPrimerBidiLimits('GGGGGGGGGG', seq))
def testLonger(self):
"""
Test a longer sequence.
"""
seq = Seq('AAAAAAAAAA'
'GGGGGGGGGG'
'AAAAAAAAAA'
'AAAAAAAAAA',
IUPAC.unambiguous_dna)
self.assertEqual((20, 40), findPrimerBidiLimits('GGGGGGGGGG', seq))
