def contamination(args):
"""
%prog contamination folder Ecoli.fasta
Remove contaminated reads. The FASTQ files in the folder will automatically
pair and filtered against Ecoli.fasta to remove contaminants using BOWTIE2.
"""
from jcvi.apps.bowtie import align
p = OptionParser(contamination.__doc__)
p.add_option("--mapped", default=False, action="store_true",
help="Retain contaminated reads instead [default: %default]")
p.set_cutoff(cutoff=800)
p.set_mateorientation(mateorientation="+-")
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(not p.print_help())
folder, ecoli = args
ecoli = get_abs_path(ecoli)
tag = "--mapped" if opts.mapped else "--unmapped"
for p, pf in iter_project(folder, 2):
align_opts = [ecoli] + p + [tag]
align_opts += ["--cutoff={0}".format(opts.cutoff), "--null"]
if opts.mateorientation:
align_opts += ["--mateorientation={0}".format(opts.mateorientation)]
samfile, logfile = align(align_opts)
def contamination(args):
"""
%prog contamination folder Ecoli.fasta
Remove contaminated reads. The FASTQ files in the folder will automatically
pair and filtered against Ecoli.fasta to remove contaminants using BOWTIE2.
"""
from jcvi.apps.bowtie import align
p = OptionParser(contamination.__doc__)
p.add_option("--mapped",
default=False,
action="store_true",
help="Retain contaminated reads instead [default: %default]")
p.set_cutoff(cutoff=800)
p.set_mateorientation(mateorientation="+-")
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(not p.print_help())
folder, ecoli = args
ecoli = get_abs_path(ecoli)
tag = "--mapped" if opts.mapped else "--unmapped"
for p, pf in iter_project(folder, 2):
align_opts = [ecoli] + p + [tag]
align_opts += ["--cutoff={0}".format(opts.cutoff), "--null"]
if opts.mateorientation:
align_opts += [
"--mateorientation={0}".format(opts.mateorientation)
]
samfile, logfile = align(align_opts)
def align(args):
"""
%prog align database.fasta read1.fq [read2.fq]
Wrapper for three modes of BWA - mem (default), aln, bwasw (long reads).
"""
valid_modes = ("bwasw", "aln", "mem")
p = OptionParser(align.__doc__)
p.add_option("--mode",
default="mem",
choices=valid_modes,
help="BWA mode [default: %default]")
p.add_option("--readtype",
choices=("pacbio", "pbread"),
help="Read type in bwa-mem")
p.set_cutoff(cutoff=800)
p.set_sam_options()
opts, args = p.parse_args(args)
mode = opts.mode
nargs = len(args)
if nargs not in (2, 3):
sys.exit(not p.print_help())
tag = "bwa-{0}: ".format(mode)
c = mem
if nargs == 2:
tag += "Single-end alignment"
if mode == "bwasw":
c = bwasw
elif mode == "aln":
c = samse
else:
assert mode != "bwasw", "Cannot use --bwasw with paired-end mode"
tag += "Paired-end alignment"
if mode == "aln":
c = sampe
logging.debug(tag)
args[0] = get_abs_path(args[0])
cmd, samfile = c(args, opts)
if cmd:
cmd = output_bam(cmd, samfile)
bam = opts.bam
unmapped = opts.unmapped
sh(cmd)
if unmapped:
dbfile, readfile = args[:2]
mopts = [samfile, "--unmapped"]
if not bam:
mopts += ["--sam"]
mapped(mopts)
FileShredder([samfile])
return samfile, None
def align(args):
"""
%prog align database.fasta read1.fq [read2.fq]
Wrapper for three modes of BWA - mem (default), aln, bwasw (long reads).
"""
valid_modes = ("bwasw", "aln", "mem")
p = OptionParser(align.__doc__)
p.add_option("--mode", default="mem", choices=valid_modes, help="BWA mode")
p.add_option("--rg", help="Read group")
p.add_option("--readtype", choices=("pacbio", "pbread", "ont2d", "intractg"),
help="Read type in bwa-mem")
p.set_cutoff(cutoff=800)
p.set_sam_options()
opts, args = p.parse_args(args)
mode = opts.mode
nargs = len(args)
if nargs not in (2, 3):
sys.exit(not p.print_help())
tag = "bwa-{0}: ".format(mode)
c = mem
if nargs == 2:
tag += "Single-end alignment"
if mode == "bwasw":
c = bwasw
elif mode == "aln":
c = samse
else:
assert mode != "bwasw", "Cannot use --bwasw with paired-end mode"
tag += "Paired-end alignment"
if mode == "aln":
c = sampe
logging.debug(tag)
cmd, samfile = c(args, opts)
if cmd:
cmd = output_bam(cmd, samfile)
bam = opts.bam
unmapped = opts.unmapped
sh(cmd)
if unmapped:
dbfile, readfile = args[:2]
mopts = [samfile, "--unmapped"]
if not bam:
mopts += ["--sam"]
mapped(mopts)
FileShredder([samfile])
return samfile, None
def align(args):
"""
%prog align database.fasta read1.fq [read2.fq]
Wrapper for `bowtie2` single-end or paired-end, depending on the number of args.
"""
from jcvi.formats.fastq import guessoffset
p = OptionParser(align.__doc__)
p.set_firstN(firstN=0)
p.add_option("--full", default=False, action="store_true", help="Enforce end-to-end alignment [default: local]")
p.add_option("--reorder", default=False, action="store_true", help="Keep the input read order [default: %default]")
p.set_cutoff(cutoff=800)
p.set_mateorientation(mateorientation="+-")
p.set_sam_options(bowtie=True)
opts, args = p.parse_args(args)
extra = opts.extra
mo = opts.mateorientation
if mo == "+-":
extra += ""
elif mo == "-+":
extra += "--rf"
else:
extra += "--ff"
PE = True
if len(args) == 2:
logging.debug("Single-end alignment")
PE = False
elif len(args) == 3:
logging.debug("Paired-end alignment")
else:
sys.exit(not p.print_help())
firstN = opts.firstN
mapped = opts.mapped
unmapped = opts.unmapped
gl = "--end-to-end" if opts.full else "--local"
dbfile, readfile = args[0:2]
dbfile = get_abs_path(dbfile)
safile = check_index(dbfile)
prefix = get_prefix(readfile, dbfile)
samfile, mapped, unmapped = get_samfile(
readfile, dbfile, bowtie=True, mapped=mapped, unmapped=unmapped, bam=opts.bam
)
logfile = prefix + ".log"
offset = guessoffset([readfile])
if not need_update(safile, samfile):
logging.error("`{0}` exists. `bowtie2` already run.".format(samfile))
return samfile, logfile
cmd = "bowtie2 -x {0}".format(dbfile)
if PE:
r1, r2 = args[1:3]
cmd += " -1 {0} -2 {1}".format(r1, r2)
cmd += " --maxins {0}".format(opts.cutoff)
mtag, utag = "--al-conc", "--un-conc"
else:
cmd += " -U {0}".format(readfile)
mtag, utag = "--al", "--un"
if mapped:
cmd += " {0} {1}".format(mtag, mapped)
if unmapped:
cmd += " {0} {1}".format(utag, unmapped)
if firstN:
cmd += " --upto {0}".format(firstN)
cmd += " -p {0}".format(opts.cpus)
cmd += " --phred{0}".format(offset)
cmd += " {0}".format(gl)
if opts.reorder:
cmd += " --reorder"
cmd += " {0}".format(extra)
# Finally the log
cmd += " 2> {0}".format(logfile)
cmd = output_bam(cmd, samfile)
sh(cmd)
print >>sys.stderr, open(logfile).read()
return samfile, logfile
def align(args):
"""
%prog align database.fasta read1.fq [read2.fq]
Wrapper for `bowtie2` single-end or paired-end, depending on the number of args.
"""
from jcvi.formats.fastq import guessoffset
p = OptionParser(align.__doc__)
p.set_firstN(firstN=0)
p.add_option("--full",
default=False,
action="store_true",
help="Enforce end-to-end alignment [default: local]")
p.add_option("--reorder",
default=False,
action="store_true",
help="Keep the input read order [default: %default]")
p.add_option("--null",
default=False,
action="store_true",
help="Do not write to SAM/BAM output")
p.add_option("--fasta",
default=False,
action="store_true",
help="Query reads are FASTA")
p.set_cutoff(cutoff=800)
p.set_mateorientation(mateorientation="+-")
p.set_sam_options(bowtie=True)
opts, args = p.parse_args(args)
extra = opts.extra
mo = opts.mateorientation
if mo == '+-':
extra += ""
elif mo == '-+':
extra += "--rf"
else:
extra += "--ff"
PE = True
if len(args) == 2:
logging.debug("Single-end alignment")
PE = False
elif len(args) == 3:
logging.debug("Paired-end alignment")
else:
sys.exit(not p.print_help())
firstN = opts.firstN
mapped = opts.mapped
unmapped = opts.unmapped
fasta = opts.fasta
gl = "--end-to-end" if opts.full else "--local"
dbfile, readfile = args[0:2]
dbfile = check_index(dbfile)
prefix = get_prefix(readfile, dbfile)
samfile, mapped, unmapped = get_samfile(readfile,
dbfile,
bowtie=True,
mapped=mapped,
unmapped=unmapped,
bam=opts.bam)
logfile = prefix + ".log"
if not fasta:
offset = guessoffset([readfile])
if not need_update(dbfile, samfile):
logging.error("`{0}` exists. `bowtie2` already run.".format(samfile))
return samfile, logfile
cmd = "bowtie2 -x {0}".format(dbfile)
if PE:
r1, r2 = args[1:3]
cmd += " -1 {0} -2 {1}".format(r1, r2)
cmd += " --maxins {0}".format(opts.cutoff)
mtag, utag = "--al-conc", "--un-conc"
else:
cmd += " -U {0}".format(readfile)
mtag, utag = "--al", "--un"
if mapped:
cmd += " {0} {1}".format(mtag, mapped)
if unmapped:
cmd += " {0} {1}".format(utag, unmapped)
if firstN:
cmd += " --upto {0}".format(firstN)
cmd += " -p {0}".format(opts.cpus)
if fasta:
cmd += " -f"
else:
cmd += " --phred{0}".format(offset)
cmd += " {0}".format(gl)
if opts.reorder:
cmd += " --reorder"
cmd += " {0}".format(extra)
# Finally the log
cmd += " 2> {0}".format(logfile)
if opts.null:
samfile = "/dev/null"
cmd = output_bam(cmd, samfile)
sh(cmd)
print(open(logfile).read(), file=sys.stderr)
return samfile, logfile
