--quality=<quality> Trimming quality [default: 20]
--adapter=<adapter> Adapter sequence [default: AGATCGGAAGAGC]
--path=<path> Path to cutadapt [default: cutadapt]
--help Output this message
"""
# Import required modules
import os
from ngs_python.fastq import fastqFind, fastqTrim
from general_python import docopt, toolbox, moab
# Extract and process arguments
args = docopt.docopt(__doc__,version = 'v1')
args['--quality'] = int(args['--quality'])
inDir, inPrefix = os.path.split(args['<inprefix>'])
toolbox.checkArg(args['--path'], 'exc')
# Extract fastq files and generate output file names
read1In, read2In = fastqFind.findFastq(prefix = inPrefix, dirList = [inDir],
pair = True, gzip = True)
read1Out = args['<outprefix>'] + '.R1.fastq.gz'
read2Out = args['<outprefix>'] + '.R2.fastq.gz'
trimLog = args['<outprefix>'] + '.log'
# Generate and submit trim command
trimCommand = fastqTrim.cutadaptTrimPaired(read1In = read1In,
read2In = read2In, read1Out = read1Out, read2Out = read2Out,
quality = args['--quality'], adapter = 'AGATCGGAAGAGC', length = 25,
path = args['--path']
)
jobID = moab.submitJob(trimCommand, stdout = trimLog, stderr = trimLog)
print jobID
path=paths['cutadapt'],
length=args['--minlength'],
quality=args['--trimqual'])
# Create QC command
fastqcCommand = fastqQC.fastQC(inFile=args['trimRead1'],
outDir=args['fastqSampleDir'],
path=paths['fastqc'])
# Empty read2 file
args['trimRead2'] = None
# Generate commands to process paired-end FASTQ files
else:
# Create trim command
trimReadsCommand = fastqTrim.cutadaptTrimPaired(read1In=args['read1'],
read2In=args['read2'],
read1Out=args['trimRead1'],
read2Out=args['trimRead2'],
path=paths['cutadapt'],
length=args['--minlength'],
quality=args['--trimqual'])
# Generate QC command
read1Fastqc = fastqQC.fastQC(inFile=args['trimRead1'],
outDir=args['fastqSampleDir'],
path=paths['fastqc'])
read2Fastqc = fastqQC.fastQC(inFile=args['trimRead2'],
outDir=args['fastqSampleDir'],
path=paths['fastqc'])
fastqcCommand = '%s && %s' % (read1Fastqc, read2Fastqc)
# Submit trimming commands
trimReadsJobID = moabJobs.add(command=trimReadsCommand,
stdout=args['trimLog'],
stderr=args['trimLog'])
# Create QC command
fastqcCommand = fastqQC.fastQC(
inFile = args['trimRead1'],
outDir = args['fastqSampleDir'],
path = paths['fastqc']
)
# Empty read2 file
args['trimRead2'] = None
# Generate commands to process paired-end FASTQ files
else:
# Create trim command
trimReadsCommand = fastqTrim.cutadaptTrimPaired(
read1In = args['read1'],
read2In = args['read2'],
read1Out = args['trimRead1'],
read2Out = args['trimRead2'],
path = paths['cutadapt'],
length = args['--minlength'],
quality = args['--trimqual']
)
# Generate QC command
read1Fastqc = fastqQC.fastQC(
inFile = args['trimRead1'],
outDir = args['fastqSampleDir'],
path = paths['fastqc']
)
read2Fastqc = fastqQC.fastQC(
inFile = args['trimRead2'],
outDir = args['fastqSampleDir'],
path = paths['fastqc']
)
"""
# Import required modules
import os
from ngs_python.fastq import fastqFind, fastqTrim
from general_python import docopt, toolbox, moab
# Extract and process arguments
args = docopt.docopt(__doc__, version='v1')
args['--quality'] = int(args['--quality'])
inDir, inPrefix = os.path.split(args['<inprefix>'])
toolbox.checkArg(args['--path'], 'exc')
# Extract fastq files and generate output file names
read1In, read2In = fastqFind.findFastq(prefix=inPrefix,
dirList=[inDir],
pair=True,
gzip=True)
read1Out = args['<outprefix>'] + '.R1.fastq.gz'
read2Out = args['<outprefix>'] + '.R2.fastq.gz'
trimLog = args['<outprefix>'] + '.log'
# Generate and submit trim command
trimCommand = fastqTrim.cutadaptTrimPaired(read1In=read1In,
read2In=read2In,
read1Out=read1Out,
read2Out=read2Out,
quality=args['--quality'],
adapter='AGATCGGAAGAGC',
length=25,
path=args['--path'])
jobID = moab.submitJob(trimCommand, stdout=trimLog, stderr=trimLog)
print jobID
args['insLog'] = args['bamPrf'] + '.insert.log'
# Generate output files for BED processing
args['bedPrf'] = os.path.join(args['smpBed'], args['name'])
args['intBed'] = args['bedPrf'] + '.initial.bed'
args['intLog'] = args['bedPrf'] + '.bam2bed.log'
args['srtBed'] = args['bedPrf'] + '.bed'
args['bedGrp'] = args['bedPrf'] + '.bedgraph'
args['bedLog'] = args['bedPrf'] + '.bedprocess.log'
# Generate object to store commands
moabJobs = moab.moabJobs()
# Create trim command and add to job dictionary
trimCommand = fastqTrim.cutadaptTrimPaired(read1In=args['read1'],
read2In=args['read2'],
read1Out=args['trmRd1'],
read2Out=args['trmRd2'],
path=paths['cutadapt'],
length=25,
quality=20,
adapter='CTGTCTCTTATACACATCT')
trimCommandID = moabJobs.add(trimCommand,
stdout=args['trmLog'],
stderr=args['trmLog'])
# Generate QC commands
read1Fastqc = fastqQC.fastQC(inFile=args['trmRd1'],
outDir=args['smpFsq'],
path=paths['fastqc'])
read2Fastqc = fastqQC.fastQC(inFile=args['trmRd2'],
outDir=args['smpFsq'],
path=paths['fastqc'])
# Combine FASTQ QC command and add to job dictionary
fastqcCommand = '%s && %s' % (read1Fastqc, read2Fastqc)
