def runSicklePaired(pe1InputFilePath: str,
pe2InputFilePath: str,
outputFolder: str = None):
if not os.path.isfile(pe1InputFilePath):
raise FileNotFoundError(
"Unable to find pe1 input file for adapter trimming at %s" %
pe1InputFilePath)
if not os.path.isfile(pe2InputFilePath):
raise FileNotFoundError(
"Unable to find pe2 input file for adapter trimming at %s" %
pe1InputFilePath)
outputFolder, inputFastq = fileHandling.fastqBasicPrep(
pe1InputFilePath, outputFolder)
pe1OutputFileName = fileHandling.stripFastqExtensions(
os.path.split(pe1InputFilePath)[1]) + ".qualTrim.fastq.gz"
pe2OutputFileName = fileHandling.stripFastqExtensions(
os.path.split(pe2InputFilePath)[1]) + ".qualTrim.fastq.gz"
singletonFileName = fileHandling.stripFastqExtensions(
os.path.split(pe1InputFilePath)[1]) + ".singleton.qualTrim.fastq.gz"
pe1OutputFilePath = os.path.join(outputFolder, pe1OutputFileName)
pe2OutputFilePath = os.path.join(outputFolder, pe2OutputFileName)
singletonOutputFilePath = os.path.join(outputFolder, singletonFileName)
sickleCommand = "%s pe -f %s -r %s -o %s -p %s -s %s -t %s -l %s -q %s -g" % (
sickleExecutable, pe1InputFilePath, pe2InputFilePath,
pe1OutputFilePath, pe2OutputFilePath, singletonOutputFilePath,
qualityEncoding, lengthThreshold, qualityThreshold)
print("RUN: %s" % sickleCommand)
exitStatus = os.system(sickleCommand)
if exitStatus != 0:
raise RuntimeError(
"Sickle command failed with non-zero exit status of %s" %
exitStatus)
return pe1OutputFilePath, pe2OutputFilePath, singletonOutputFilePath
def runScythe(inputFilePath: str,
outputFolder: str = None,
adaptersFilePath: str = None):
if not os.path.isfile(inputFilePath):
raise FileNotFoundError(
"Unable to find input file for adapter trimming at %s" %
inputFilePath)
if not adaptersFilePath:
return False
if not os.path.isfile(adaptersFilePath):
raise FileNotFoundError("Unable to find adapters file at %s" %
adaptersFilePath)
outputFolder, inputFastq = fileHandling.fastqBasicPrep(
inputFilePath, outputFolder)
outputFileName = fileHandling.stripFastqExtensions(
inputFastq) + ".adapterTrim.fastq"
outputFilePath = os.path.join(outputFolder, outputFileName)
scytheCommand = "%s -a %s -o %s %s" % (scytheExecutable, adaptersFilePath,
outputFilePath, inputFilePath)
print("RUN: %s" % scytheCommand)
exitStatus = os.system(scytheCommand)
if exitStatus != 0:
raise RuntimeError(
"Scythe command failed with non-zero exit status of %s" %
exitStatus)
return outputFilePath
def runCutadapt(inputFilePath:str, outputFolder:str=None, adaptersFilePath:str=None):
if not os.path.isfile(inputFilePath):
raise FileNotFoundError("Unable to find input file for adapter trimming at %s" % inputFilePath)
if not adaptersFilePath:
return False
if not os.path.isfile(adaptersFilePath):
raise FileNotFoundError("Unable to find adapters file at %s" %adaptersFilePath)
outputFolder, inputFastq = fileHandling.fastqBasicPrep(inputFilePath, outputFolder)
outputFileName = fileHandling.stripFastqExtensions(inputFastq) + ".adapterTrim.fastq.gz"
outputFilePath = os.path.join(outputFolder, outputFileName)
cutadaptCommand = "conda activate %s; cutadapt -j %s -a FILE:%s -o %s %s; cutadapt conda deactivate" %(cutadaptEnvironmentName, cpuCount, adaptersFilePath, outputFilePath, inputFilePath)
print("RUN: %s" %cutadaptCommand)
exitStatus = os.system(cutadaptCommand)
if exitStatus != 0:
raise RuntimeError("Scythe command failed with non-zero exit status of %s" %exitStatus)
return outputFilePath
def bwaAlignAndCompress(
pe1Reads: str,
pe2Reads: str = "",
outputFilePath: str = None,
referenceGenome: str = referenceGenomeEnv
): #keep pe2 file defaulting to empty string for proper BWA command
if not outputFilePath:
outputFilePath = fileHandling.stripFastqExtensions(pe1Reads) + ".bam"
bwaCommand = "%s mem -t %s %s %s %s" % (
bwaExecutable, cpuCount, referenceGenome, pe1Reads, pe2Reads)
samtoolsBAMCommand = "%s view -@%s -bS - > %s" % (samtoolsExecutable,
cpuCount, outputFilePath)
fullCommand = "%s | %s" % (bwaCommand, samtoolsBAMCommand)
print("RUN: %s" % fullCommand)
exitStatus = os.system(fullCommand)
if exitStatus != 0:
raise RuntimeError(
"Alignment and compression command returned a non-zero exit status."
)
return outputFilePath