def executeFastqcCommand(rawReadsLocation, outputDir):
# Determine the location of executable
xmlFile = 'configs/executables.xml'
nodeName = 'FASTQC'
fastqc_path = rx.extractLocationFromExecutables(xmlFile,nodeName)
# create output dir if doesn't exist
if not os.path.exists(outputDir):
os.makedirs(outputDir)
logging.info('Directory created at ' + outputDir)
# create fastqc command and execute
for readFile in rawReadsLocation:
fastqc_command = [fastqc_path,"-o", outputDir, readFile]
logging.info('Fastqc command - ' + ' '.join(fastqc_command))
subprocess.call(fastqc_command)
def call_kmergenie(dataLocation, outputDir):
'''
The function calls kmer-genie and captures screen output in report.txt file
:param dataLocation: Location containing .fq/.fastq files
:param outputDir: Location where the output will be written
'''
# Determine the location of executable
xmlFile = 'configs/executables.xml'
nodeName = 'KMERGENIE'
kmergenie_path = rx.extractLocationFromExecutables(xmlFile,nodeName)
input_parameterFile = prepare_inputFile(dataLocation, outputDir)
if input_parameterFile is None:
logging.error('Error creating kmer-genie parameter file')
exit(0)
# Need to change dir as the tool doesn't provide option to specify output path
os.chdir(outputDir)
# Open file for recording kmer-genie output
reportFile = os.path.join(outputDir,'report.txt')
report = open(reportFile, mode='w')
kmergenie_command = [kmergenie_path, input_parameterFile]
print('kmergenie command - ' + ' '.join(kmergenie_command))
logging.info('kmergenie: kmergenie command - ' + ' '.join(kmergenie_command))
p = subprocess.Popen(kmergenie_command, stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
for line in p.stdout:
report.write(str(line))
p.stdout.flush()
report.close()
return 0
def executeFastqTrim_sickle(dataLocation, se_or_pe, outputDir, quality_type = 'illumina', quality_threhold = 20, *pairedFiles):
'''
:param dataLocation: Folder containing input data as .fq or .fasta
:param se_or_pe: 'S' or 'P'
:param outputDir:
:param quality_type: 'illumina', 'sanger'or 'solexa'
:param quality_threhold: integer value
param pairedFiles: Pairing of files as specified by user in a dictionary format {'ForwardReads.fq:ReverseReads.fq'}.
'''
# Determine the location of executable
xmlFile = 'configs/executables.xml'
nodeName = 'SICKLE'
tool_path = rx.extractLocationFromExecutables(xmlFile,nodeName)
dataType = ['S','P']
if se_or_pe not in dataType:
logging.error('executeFastqTrim_sickle: Incorrect specification for read data type. Specify S or P, opposed to the provided option - ' + se_or_pe)
return None
# create output dir if doesn't exist
if not os.path.exists(outputDir):
os.makedirs(outputDir)
logging.info('executeFastqTrim_sickle: Directory created at ' + outputDir)
if quality_type not in ['illumina', 'sanger', 'solexa']:
logging.error('Quality type must be one of these : illumina, sanger, solexa')
return None
if se_or_pe == 'S':
# expect a folder with .fq
fq_files = utilities.get_fastq_files(dataLocation)
if fq_files is None:
logging.error('No FASTQ files found.')
# Open file for recording cutadapt output
reportFile = os.path.join(outputDir,'sickleReport.txt')
report = open(reportFile, mode='w')
for fq_file in fq_files:
(dirname, filename) = os.path.split(fq_file)
outPath = os.path.join(outputDir, filename)
sickle_command = [tool_path, 'se', '-f', fq_file, '-t', quality_type, '-q', str(quality_threhold),'-o', outPath]
print('sickle command - ' + ' '.join(sickle_command))
logging.info('executeFastqTrim_sickle: sickle command - ' + ' '.join(sickle_command))
p = subprocess.Popen(sickle_command, stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
for line in p.stdout:
report.write(str(line))
p.stdout.flush()
report.close()
return 0
else:
# expect a folder and pairing of files. Files should be paired in same way as the adapters are specified.
pairedFiles_dict = pairedFiles[0]
if not isinstance(pairedFiles_dict, dict):
logging.error('executeFastqTrim_sickle: requires dictionary format for using the Paired-end trimming')
return None
# Open file for recording cutadapt output
reportFile = os.path.join(outputDir,'sickleReport.txt')
report = open(reportFile, mode='w')
strand1_files = pairedFiles_dict.keys()
for st1_file in strand1_files:
st2_file = pairedFiles_dict[st1_file]
# Create full path for input files
st1_file_path = os.path.join(dataLocation,st1_file)
st2_file_path = os.path.join(dataLocation,st2_file)
if os.path.exists(st1_file_path) & os.path.exists(st2_file_path):
st1_outPath = os.path.join(outputDir, st1_file)
st2_outPath = os.path.join(outputDir, st2_file)
singles_outPath = os.path.join(outputDir,'trimmedSingles.fastq')
sickle_command = [tool_path, 'pe', '-f', st1_file_path,'-r',st2_file_path, \
'-t', quality_type, '-q', str(quality_threhold), \
'-o', st1_outPath, '-p', st2_outPath, '-s',singles_outPath]
print('sickle command - ' + ' '.join(sickle_command))
logging.info('executeFastqTrim_sickle: sickle command - ' + ' '.join(sickle_command))
p = subprocess.Popen(sickle_command, stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
for line in p.stdout:
report.write(str(line))
p.stdout.flush()
else: # Return with error message if file(s) don't exist
msg = 'Check if files exist: ' + st1_file_path + ',' + st2_file_path
logging.error(msg)
return None
report.close()
return 0
def removeAdaptors_cutadapt(dataLocation, se_or_pe, adapterDict, fixBaseNumber, outputDir, *pairedFiles):
'''
:param adapterDict: Adaptar dictionary in form compatible with "cutadapt". Like [ADAPTER:-a] for a 3' adapter type,
[ADAPTER:-g] for a 5' adapter,[ADAPTER$:-a] for anchored 3' adapter,[^ADAPTER:-g] for anchored 5' adapter, and
[ADAPTER:b] for both 5' and 3' adapter type. Similarly, it can take -A, -B and -G for paired versions.
Details - "http://cutadapt.readthedocs.org/en/stable/guide.html#removing-adapters"
:param se_or_pe: Single or paired end data. Options - 'S' or 'P'
:param fixBaseNumber: Remove fixed number of bases. -u option like -u 5 (remove first 5 bases)
:param *pairedFiles: Pairing of files as specified by user in a dictionary format {'ForwardReads.fq:ReverseReads.fq'}.
'''
dataType = ['S','P']
# Determine the location of executable
xmlFile = 'configs/executables.xml'
nodeName = 'CUTADAPT'
tool_path = rx.extractLocationFromExecutables(xmlFile,nodeName)
if se_or_pe not in dataType:
logging.error('removeAdaptors_cutadapt: Incorrect specification for read data type. Specify S or P, opposed to the provided option - ' + se_or_pe)
return None
# create output dir if doesn't exist
if not os.path.exists(outputDir):
os.makedirs(outputDir)
logging.info('removeAdaptors_cutadapt: Directory created at ' + outputDir)
# check if correct adapter types specified
allowedAdapterTypes = ['-a', '-g', '-b', '-A', '-G', '-B'] # As specified by cutadapt
userProvidedAdapterTypes = adapterDict.values()
for type in userProvidedAdapterTypes:
if type not in allowedAdapterTypes:
logging.error('removeAdaptors_cutadapt: Wrong keys provided for specifying adapters. Wrong entry found is : ' + type)
return None
# Arrange adapter list to a string
adapters = adapterDict.keys()
adapter_parameter_format = ''
for adapter in adapters:
adapter_parameter_format = adapter_parameter_format + adapterDict[adapter] + ' ' + adapter + ' '
print('Adapters and parameters :' + adapter_parameter_format)
logging.info('removeAdaptors_cutadapt: Adapters and parameters :' + adapter_parameter_format)
if se_or_pe == 'S':
# expect a folder with .fq
fq_files = utilities.get_fastq_files(dataLocation)
if fq_files is None:
logging.error('No FASTQ files found.')
# Open file for recording cutadapt output
reportFile = os.path.join(outputDir,'cutadaptReport.txt')
report = open(reportFile, mode='w')
for fq_file in fq_files:
(dirname, filename) = os.path.split(fq_file)
outPath = os.path.join(outputDir, filename)
cutadapt_command = [tool_path, adapter_parameter_format, '-u',fixBaseNumber,'-o', outPath,fq_file]
print('cutadapt command - ' + ' '.join(cutadapt_command))
logging.info('removeAdaptors_cutadapt: cutadapt command - ' + ' '.join(cutadapt_command))
p = subprocess.Popen(cutadapt_command, stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
for line in p.stdout:
report.write(str(line))
p.stdout.flush()
report.close()
return 0
else:
# expect a folder and pairing of files. Files should be paired in same way as the adapters are specified.
pairedFiles_dict = pairedFiles[0]
if not isinstance(pairedFiles_dict, dict):
logging.error('removeAdaptors_cutadapt: requires dictionary format for using the Paired-end trimming')
return None
# Open file for recording cutadapt output
reportFile = os.path.join(outputDir,'cutadaptReport.txt')
report = open(reportFile, mode='w')
strand1_files = pairedFiles_dict.keys()
for st1_file in strand1_files:
st2_file = pairedFiles_dict[st1_file]
# Create full path for input files
st1_file_path = os.path.join(dataLocation,st1_file)
st2_file_path = os.path.join(dataLocation,st2_file)
if os.path.exists(st1_file_path) & os.path.exists(st2_file_path):
st1_outPath = os.path.join(outputDir, st1_file)
st2_outPath = os.path.join(outputDir, st2_file)
cutadapt_command = [tool_path, adapter_parameter_format, '-u',fixBaseNumber,'-o', st1_outPath, '-p',st2_outPath, st1_file_path, st2_file_path]
print('cutadapt command - ' + ' '.join(cutadapt_command))
logging.info('removeAdaptors_cutadapt: cutadapt command - ' + ' '.join(cutadapt_command))
p = subprocess.Popen(cutadapt_command, stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
for line in p.stdout:
report.write(str(line))
p.stdout.flush()
else: # Return with error message if file(s) don't exist
msg = 'Check if files exist: ' + st1_file_path + ',' + st2_file_path
logging.error(msg)
return None
report.close()
return 0