def testCactusRealign(self):
"""Runs cactus realign using the default parameters and checks that the realigned output cigars align
the same subsequences.
"""
for seqFile1, seqFile2 in seqFilePairGenerator():
lastzOutput = getTempFile(rootDir=self.tempDir)
runLastz(seqFile1,
seqFile2,
alignmentsFile=lastzOutput,
lastzArguments=self.defaultLastzArguments,
work_dir=self.tempDir)
realignOutput = getTempFile(rootDir=self.tempDir)
runCactusRealign(seqFile1,
seqFile2,
inputAlignmentsFile=lastzOutput,
outputAlignmentsFile=realignOutput,
realignArguments=self.defaultRealignArguments,
work_dir=self.tempDir)
for realignLine, lastzLine in zip(
[i for i in open(lastzOutput, 'r') if i != ''],
[i for i in open(realignOutput, 'r') if i != '']):
realignCigar = cigarReadFromString(realignLine)
lastzCigar = cigarReadFromString(lastzLine)
self.assertTrue(realignCigar.sameCoordinates(lastzCigar))
def testCactusRealignDummy(self):
"""Runs cactus realign using the "rescoreOriginalAlignment" mode
and checks the output is equivalent to what you'd get by just running lastz.
"""
for seqFile1, seqFile2 in seqFilePairGenerator():
lastzOutput = getTempFile(rootDir=self.tempDir)
runLastz(seqFile1,
seqFile2,
alignmentsFile=lastzOutput,
lastzArguments=self.defaultLastzArguments)
realignOutput = getTempFile(rootDir=self.tempDir)
runCactusRealign(seqFile1,
seqFile2,
inputAlignmentsFile=lastzOutput,
outputAlignmentsFile=realignOutput,
realignArguments=self.defaultRealignArguments +
" --rescoreOriginalAlignment")
for realignLine, lastzLine in zip(
[i for i in open(lastzOutput, 'r') if i != ''],
[i for i in open(realignOutput, 'r') if i != '']):
realignCigar = cigarReadFromString(realignLine)
lastzCigar = cigarReadFromString(lastzLine)
self.assertTrue(realignCigar != None)
self.assertTrue(realignCigar == lastzCigar)
def testCactusRealignRescoreByIdentityAndProb(self):
"""Runs cactus realign using the default parameters and checks that the realigned output cigars align
the same subsequences.
"""
for seqFile1, seqFile2 in seqFilePairGenerator():
lastzOutput = getTempFile(rootDir=self.tempDir)
runLastz(seqFile1,
seqFile2,
alignmentsFile=lastzOutput,
lastzArguments=self.defaultLastzArguments,
work_dir=self.tempDir)
realignByIdentityOutput = getTempFile(rootDir=self.tempDir)
runCactusRealign(seqFile1,
seqFile2,
inputAlignmentsFile=lastzOutput,
outputAlignmentsFile=realignByIdentityOutput,
realignArguments=self.defaultRealignArguments +
" --rescoreByIdentity",
work_dir=self.tempDir)
realignByPosteriorProbOutput = getTempFile(rootDir=self.tempDir)
runCactusRealign(seqFile1,
seqFile2,
inputAlignmentsFile=lastzOutput,
outputAlignmentsFile=realignByPosteriorProbOutput,
realignArguments=self.defaultRealignArguments +
" --rescoreByPosteriorProb",
work_dir=self.tempDir)
realignByIdentityIgnoringGapsOutput = getTempFile(
rootDir=self.tempDir)
runCactusRealign(
seqFile1,
seqFile2,
inputAlignmentsFile=lastzOutput,
outputAlignmentsFile=realignByIdentityIgnoringGapsOutput,
realignArguments=self.defaultRealignArguments +
" --rescoreByIdentityIgnoringGaps",
work_dir=self.tempDir)
for realignLineByIdentity, realignLineByPosteriorProb, realignLineByIdentityIgnoringGaps, lastzLine in \
zip([ i for i in open(realignByIdentityOutput, 'r') if i != '' ], \
[ i for i in open(realignByPosteriorProbOutput, 'r') if i != '' ], \
[ i for i in open(realignByIdentityIgnoringGapsOutput, 'r') if i != '' ], \
[ i for i in open(lastzOutput, 'r') if i != '' ]):
realignCigarByIdentity = cigarReadFromString(
realignLineByIdentity)
realignCigarByPosteriorProb = cigarReadFromString(
realignLineByPosteriorProb)
realignCigarByIdentityIgnoringGaps = cigarReadFromString(
realignLineByIdentityIgnoringGaps)
lastzCigar = cigarReadFromString(lastzLine)
#Check scores are as expected
self.assertTrue(realignCigarByIdentity.score >= 0)
self.assertTrue(realignCigarByIdentity.score <= 100.0)
self.assertTrue(realignCigarByPosteriorProb.score >= 0)
self.assertTrue(realignCigarByPosteriorProb.score <= 100.0)
self.assertTrue(realignCigarByIdentityIgnoringGaps.score >= 0)
self.assertTrue(
realignCigarByIdentityIgnoringGaps.score <= 100.0)
def testCPecanRealign(self):
"""Runs cPecanRealign using the default parameters and checks that the realigned output cigars align
the same subsequences.
"""
for seqFile1, seqFile2 in seqFilePairGenerator():
realignCommand, lastzCommand = getCommands(seqFile1, seqFile2)
for realignLine, lastzLine in zip([ i for i in popenCatch(realignCommand).split("\n") if i != '' ],
[ i for i in popenCatch(lastzCommand).split("\n") if i != '' ]):
realignCigar = cigarReadFromString(realignLine)
lastzCigar = cigarReadFromString(lastzLine)
self.assertTrue(realignCigar.sameCoordinates(lastzCigar))
def testCPecanRealignDummy(self):
"""Runs cPecanRealign using the "rescoreOriginalAlignment" mode
and checks the output is equivalent to what you'd get by just running lastz.
"""
for seqFile1, seqFile2 in seqFilePairGenerator():
realignCommand, lastzCommand = getCommands(seqFile1, seqFile2, "--rescoreOriginalAlignment")
for realignLine, lastzLine in zip([ i for i in popenCatch(realignCommand).split("\n") if i != '' ],
[ i for i in popenCatch(lastzCommand).split("\n") if i != '' ]):
realignCigar = cigarReadFromString(realignLine)
lastzCigar = cigarReadFromString(lastzLine)
self.assertTrue(realignCigar != None)
self.assertTrue(realignCigar == lastzCigar)
def testCactusRealign(self):
"""Runs cactus realign using the default parameters and checks that the realigned output cigars align
the same subsequences.
"""
for seqFile1, seqFile2 in seqFilePairGenerator():
lastzOutput = getTempFile(rootDir=self.tempDir)
runLastz(seqFile1, seqFile2, alignmentsFile=lastzOutput,
lastzArguments=self.defaultLastzArguments)
realignOutput = getTempFile(rootDir=self.tempDir)
runCactusRealign(seqFile1, seqFile2, inputAlignmentsFile = lastzOutput,
outputAlignmentsFile = realignOutput,
realignArguments=self.defaultRealignArguments)
for realignLine, lastzLine in zip([ i for i in open(lastzOutput, 'r') if i != '' ],
[ i for i in open(realignOutput, 'r') if i != '' ]):
realignCigar = cigarReadFromString(realignLine)
lastzCigar = cigarReadFromString(lastzLine)
self.assertTrue(realignCigar.sameCoordinates(lastzCigar))
def testCactusRealignDummy(self):
"""Runs cactus realign using the "rescoreOriginalAlignment" mode
and checks the output is equivalent to what you'd get by just running lastz.
"""
for seqFile1, seqFile2 in seqFilePairGenerator():
lastzOutput = getTempFile(rootDir=self.tempDir)
runLastz(seqFile1, seqFile2, alignmentsFile=lastzOutput,
lastzArguments=self.defaultLastzArguments)
realignOutput = getTempFile(rootDir=self.tempDir)
runCactusRealign(seqFile1, seqFile2, inputAlignmentsFile = lastzOutput,
outputAlignmentsFile = realignOutput,
realignArguments=self.defaultRealignArguments + " --rescoreOriginalAlignment")
for realignLine, lastzLine in zip([ i for i in open(lastzOutput, 'r') if i != '' ],
[ i for i in open(realignOutput, 'r') if i != '' ]):
realignCigar = cigarReadFromString(realignLine)
lastzCigar = cigarReadFromString(lastzLine)
self.assertTrue(realignCigar != None)
self.assertTrue(realignCigar == lastzCigar)
def testCactusRealignRescoreByIdentityAndProb(self):
"""Runs cactus realign using the default parameters and checks that the realigned output cigars align
the same subsequences.
"""
for seqFile1, seqFile2 in seqFilePairGenerator():
lastzOutput = getTempFile(rootDir=self.tempDir)
runLastz(seqFile1, seqFile2, alignmentsFile=lastzOutput,
lastzArguments=self.defaultLastzArguments)
realignByIdentityOutput = getTempFile(rootDir=self.tempDir)
runCactusRealign(seqFile1, seqFile2, inputAlignmentsFile=lastzOutput,
outputAlignmentsFile=realignByIdentityOutput,
realignArguments=self.defaultRealignArguments + " --rescoreByIdentity")
realignByPosteriorProbOutput = getTempFile(rootDir=self.tempDir)
runCactusRealign(seqFile1, seqFile2, inputAlignmentsFile=lastzOutput,
outputAlignmentsFile=realignByPosteriorProbOutput,
realignArguments=self.defaultRealignArguments + " --rescoreByPosteriorProb")
realignByIdentityIgnoringGapsOutput = getTempFile(rootDir=self.tempDir)
runCactusRealign(seqFile1, seqFile2, inputAlignmentsFile=lastzOutput,
outputAlignmentsFile=realignByIdentityIgnoringGapsOutput,
realignArguments=self.defaultRealignArguments + " --rescoreByIdentityIgnoringGaps")
for realignLineByIdentity, realignLineByPosteriorProb, realignLineByIdentityIgnoringGaps, lastzLine in \
zip([ i for i in open(realignByIdentityOutput, 'r') if i != '' ], \
[ i for i in open(realignByPosteriorProbOutput, 'r') if i != '' ], \
[ i for i in open(realignByIdentityIgnoringGapsOutput, 'r') if i != '' ], \
[ i for i in open(lastzOutput, 'r') if i != '' ]):
realignCigarByIdentity = cigarReadFromString(realignLineByIdentity)
realignCigarByPosteriorProb = cigarReadFromString(realignLineByPosteriorProb)
realignCigarByIdentityIgnoringGaps = cigarReadFromString(realignLineByIdentityIgnoringGaps)
lastzCigar = cigarReadFromString(lastzLine)
#Check scores are as expected
self.assertTrue(realignCigarByIdentity.score >= 0)
self.assertTrue(realignCigarByIdentity.score <= 100.0)
self.assertTrue(realignCigarByPosteriorProb.score >= 0)
self.assertTrue(realignCigarByPosteriorProb.score <= 100.0)
self.assertTrue(realignCigarByIdentityIgnoringGaps.score >= 0)
self.assertTrue(realignCigarByIdentityIgnoringGaps.score <= 100.0)
def testCPecanRealignRescoreByIdentityAndProb(self):
"""Runs cactus realign using the default parameters and checks that the realigned output cigars align
the same subsequences.
"""
for seqFile1, seqFile2 in seqFilePairGenerator():
realignCommandByIdentity, lastzCommand = getCommands(seqFile1, seqFile2, realignArguments="--rescoreByIdentity")
realignCommandByPosteriorProb = getCommands(seqFile1, seqFile2, realignArguments="--rescoreByPosteriorProb")[0]
realignCommandByIdentityIgnoringGaps = getCommands(seqFile1, seqFile2, realignArguments="--rescoreByIdentityIgnoringGaps")[0]
for realignLineByIdentity, realignLineByPosteriorProb, realignLineByIdentityIgnoringGaps, lastzLine in \
zip([ i for i in popenCatch(realignCommandByIdentity).split("\n") if i != '' ], \
[ i for i in popenCatch(realignCommandByPosteriorProb).split("\n") if i != '' ], \
[ i for i in popenCatch(realignCommandByIdentityIgnoringGaps).split("\n") if i != '' ], \
[ i for i in popenCatch(lastzCommand).split("\n") if i != '' ]):
realignCigarByIdentity = cigarReadFromString(realignLineByIdentity)
realignCigarByPosteriorProb = cigarReadFromString(realignLineByPosteriorProb)
realignCigarByIdentityIgnoringGaps = cigarReadFromString(realignLineByIdentityIgnoringGaps)
lastzCigar = cigarReadFromString(lastzLine)
#Check scores are as expected
self.assertTrue(realignCigarByIdentity.score >= 0)
self.assertTrue(realignCigarByIdentity.score <= 100.0)
self.assertTrue(realignCigarByPosteriorProb.score >= 0)
self.assertTrue(realignCigarByPosteriorProb.score <= 100.0)
self.assertTrue(realignCigarByIdentityIgnoringGaps.score >= 0)
self.assertTrue(realignCigarByIdentityIgnoringGaps.score <= 100.0)
