def test_extractRead(self):
bowtie2Wrap.extractRead(self.targetAlignFile, self.fastqOutFile)
expectedReadId = [
"HWI-ST998R:270:H7NJ9ADXX:1:1101:1797:1927",
"HWI-ST998R:270:H7NJ9ADXX:1:1101:1797:1927:A"
]
with open(self.fastqOutFile, 'r') as fp:
count = 0
for r in seqParse.parse(fp, 'fastq'):
self.assertTrue(
count < len(expectedReadId),
"Extract Read: Expected number of reads Mismatch!")
self.assertTrue(r.id == expectedReadId[count],
"Extract Read: Expected Reads Mismatch!")
count += 1
self.assertTrue(
count == len(expectedReadId),
"Extract Read: Expected number of reads Mismatch!")
def processPathoMap(pathoMapOptions):
procPathoMapOptions = copyPathoMapOptions(pathoMapOptions)
# Splitting reference files if bigger than MAX_REF_FILE_SIZE
ptargetRefFiles = []
for filePath in pathoMapOptions.targetRefFiles:
if pathoMapOptions.verbose:
print "Checking whether the file: " + filePath + " needs to be split"
files = splitCheck(filePath, pathoMapOptions.MAX_REF_FILE_SIZE)
for f in files:
ptargetRefFiles.append(f)
procPathoMapOptions.targetRefFiles = ptargetRefFiles
pfilterRefFiles = []
for filePath in pathoMapOptions.filterRefFiles:
if pathoMapOptions.verbose:
print "Checking whether the file: " + filePath + " needs to be split"
files = splitCheck(filePath, pathoMapOptions.MAX_REF_FILE_SIZE)
for f in files:
pfilterRefFiles.append(f)
procPathoMapOptions.filterRefFiles = pfilterRefFiles
# Creating Index if it does not exist
bowtie2Options = bowtie2Wrap.Bowtie2Options()
bowtie2Options.verbose = procPathoMapOptions.verbose
bowtie2Options.btHome = procPathoMapOptions.btHome
bowtie2Options.indexDir = procPathoMapOptions.indexDir
for filePath in ptargetRefFiles:
bowtie2Options.refFile = filePath
(_, tail) = os.path.split(filePath)
(base, _) = os.path.splitext(tail)
bowtie2Options.btIndexPrefix = base
if pathoMapOptions.verbose:
print "Creating bowtie2 index for: " + filePath
bowtie2Wrap.create_bowtie2_index(bowtie2Options)
procPathoMapOptions.targetIndexPrefixes.append(base)
for filePath in pfilterRefFiles:
bowtie2Options.refFile = filePath
(_, tail) = os.path.split(filePath)
(base, _) = os.path.splitext(tail)
bowtie2Options.btIndexPrefix = base
if pathoMapOptions.verbose:
print "Creating bowtie2 index for: " + filePath
bowtie2Wrap.create_bowtie2_index(bowtie2Options)
procPathoMapOptions.filterIndexPrefixes.append(base)
# Creating the Alignment file
bowtie2Options = bowtie2Wrap.Bowtie2Options()
bowtie2Options.verbose = procPathoMapOptions.verbose
bowtie2Options.btHome = procPathoMapOptions.btHome
bowtie2Options.numThreads = procPathoMapOptions.numThreads
bowtie2Options.outDir = procPathoMapOptions.outDir
bowtie2Options.indexDir = procPathoMapOptions.indexDir
bowtie2Options.readFile = procPathoMapOptions.inReadFile
bowtie2Options.readFilePair1 = procPathoMapOptions.inReadFilePair1
bowtie2Options.readFilePair2 = procPathoMapOptions.inReadFilePair2
if (len(procPathoMapOptions.inReadFilePair1) > 0
and len(procPathoMapOptions.inReadFilePair2) > 0):
bowtie2Options.pairedReadFlag = True
if procPathoMapOptions.targetAlignParameters is not None:
bowtie2Options.additionalOptions = procPathoMapOptions.targetAlignParameters
for indexPrefix in procPathoMapOptions.targetIndexPrefixes:
bowtie2Options.btIndexPrefix = procPathoMapOptions.indexDir + os.sep + indexPrefix
bowtie2Options.outAlignFile = procPathoMapOptions.exp_tag + indexPrefix + ".sam"
if pathoMapOptions.verbose:
print "Creating bowtie2 alignment: " + bowtie2Options.outAlignFile
bowtie2Wrap.run_bowtie2(bowtie2Options)
procPathoMapOptions.targetAlignFiles.append(
procPathoMapOptions.outDir + os.sep + bowtie2Options.outAlignFile)
# Appending the Alignment files and Filtering
if len(procPathoMapOptions.targetAlignFiles) > 1:
appendAlignFile = procPathoMapOptions.outDir + os.sep + procPathoMapOptions.exp_tag + "appendAlign.sam"
if pathoMapOptions.verbose:
print "Appending alignment files to: " + appendAlignFile
append_file(appendAlignFile, procPathoMapOptions.targetAlignFiles)
else:
appendAlignFile = procPathoMapOptions.targetAlignFiles[0]
if len(procPathoMapOptions.filterIndexPrefixes) > 0:
bowtie2Options.readFile = procPathoMapOptions.outDir + os.sep + procPathoMapOptions.exp_tag + "appendAlign.fq"
bowtie2Options.readFilePair1 = ""
bowtie2Options.readFilePair2 = ""
bowtie2Options.pairedReadFlag = False
if procPathoMapOptions.filterAlignParameters is not None:
bowtie2Options.additionalOptions = procPathoMapOptions.filterAlignParameters
bowtie2Wrap.extractRead(appendAlignFile, bowtie2Options.readFile)
for indexPrefix in procPathoMapOptions.filterIndexPrefixes:
bowtie2Options.btIndexPrefix = procPathoMapOptions.indexDir + os.sep + indexPrefix
bowtie2Options.outAlignFile = procPathoMapOptions.exp_tag + indexPrefix + ".sam"
if pathoMapOptions.verbose:
print "Creating bowtie2 alignment: " + bowtie2Options.outAlignFile
bowtie2Wrap.run_bowtie2(bowtie2Options)
procPathoMapOptions.filterAlignFiles.append(
procPathoMapOptions.outDir + os.sep +
bowtie2Options.outAlignFile)
# Filtering the Alignment file
outAlignFile = procPathoMapOptions.outDir + os.sep + procPathoMapOptions.outAlignFile
if pathoMapOptions.verbose:
print "Filtering and creating the alignment: " + outAlignFile
if len(procPathoMapOptions.filterAlignFiles) > 0:
filter_alignment(appendAlignFile, procPathoMapOptions.filterAlignFiles,
outAlignFile)
elif ((len(procPathoMapOptions.targetAlignFiles) > 1) or \
(len(procPathoMapOptions.targetIndexPrefixes) > 0)):
os.rename(appendAlignFile, outAlignFile)
else: # Input appendAlignFile provided by user, hence make a copy for outAlignFile
shutil.copy(appendAlignFile, outAlignFile)
