def testFastaReadWriteC(self):
"""Tests consistency with C version of this function.
"""
tempFile = getTempFile()
self.tempFiles.append(tempFile)
tempFile2 = getTempFile()
self.tempFiles.append(tempFile2)
for test in xrange(0, self.testNo):
fastaNumber = random.choice(xrange(10))
l = [ getRandomSequence() for i in xrange(fastaNumber) ]
fileHandle = open(tempFile, 'w')
for name, seq in l:
fastaWrite(fileHandle, name, seq)
fileHandle.close()
command = "sonLib_fastaCTest %s %s" % (tempFile, tempFile2)
print command
system(command)
fileHandle = open(tempFile2, 'r')
l.reverse()
for i in fastaRead(fileHandle):
name, seq = i
assert i == l.pop()
fastaWrite(sys.stdout, name, seq)
fileHandle.close()
def main():
parser = argparse.ArgumentParser()
parser.add_argument("--unmasked")
parser.add_argument("--hardmasked")
parser.add_argument("--softmasked")
options = parser.parse_args()
unmasked = open(options.unmasked, 'r')
hardmasked = open(options.hardmasked, 'r')
softmasked = open(options.softmasked, 'w')
for unmaskedInfo, hardmaskedInfo in itertools.izip(bioio.fastaRead(unmasked), bioio.fastaRead(hardmasked)):
unmaskedName, unmaskedSeq = unmaskedInfo
hardmaskedName, hardmaskedSeq = hardmaskedInfo
print("processing sequence: %s" % unmaskedName)
assert unmaskedName == hardmaskedName
softmaskedSeq = ''
for unmaskedBase, hardmaskedBase in zip(unmaskedSeq, hardmaskedSeq):
if unmaskedBase.upper() == hardmaskedBase.upper():
softmaskedBase = unmaskedBase.upper()
else:
if not hardmaskedBase.upper() == 'N':
print("Warning: encountered differing base: %s %s" % (unmaskedBase, hardmaskedBase))
softmaskedBase = unmaskedBase.lower()
softmaskedSeq += softmaskedBase
bioio.fastaWrite(softmasked, unmaskedName, softmaskedSeq)
def testFastaReadWrite(self):
tempFile = getTempFile()
self.tempFiles.append(tempFile)
for test in xrange(0, self.testNo):
fastaNumber = random.choice(xrange(10))
l = [ getRandomSequence() for i in xrange(fastaNumber) ]
fileHandle = open(tempFile, 'w')
for name, seq in l:
fastaWrite(fileHandle, name, seq)
fileHandle.close()
fileHandle = open(tempFile, 'r')
l.reverse()
for i in fastaRead(fileHandle):
assert i == l.pop()
name, seq = i
fastaWrite(sys.stdout, name, seq)
fileHandle.close()