def sampe(args, opts):
"""
%prog sampe database.fasta read1.fq read2.fq
Wrapper for `bwa sampe`. Output will be read1.sam.
"""
dbfile, read1file, read2file = args
dbfile = check_index(dbfile)
sai1file = check_aln(dbfile, read1file, cpus=opts.cpus)
sai2file = check_aln(dbfile, read2file, cpus=opts.cpus)
samfile, _, unmapped = get_samfile(read1file,
dbfile,
bam=opts.bam,
unmapped=opts.unmapped)
if not need_update((dbfile, sai1file, sai2file), samfile):
logging.error("`{0}` exists. `bwa samse` already run.".format(samfile))
return "", samfile
cmd = "bwa sampe " + " ".join(
(dbfile, sai1file, sai2file, read1file, read2file))
cmd += " " + opts.extra
if opts.cutoff:
cmd += " -a {0}".format(opts.cutoff)
if opts.uniq:
cmd += " -n 1"
return cmd, samfile
def sampe(args, opts):
"""
%prog sampe database.fasta read1.fq read2.fq
Wrapper for `bwa sampe`. Output will be read1.sam.
"""
dbfile, read1file, read2file = args
safile = check_index(dbfile)
sai1file = check_aln(dbfile, read1file, cpus=opts.cpus)
sai2file = check_aln(dbfile, read2file, cpus=opts.cpus)
samfile, _, unmapped = get_samfile(read1file, dbfile,
bam=opts.bam, unmapped=opts.unmapped)
if not need_update((safile, sai1file, sai2file), samfile):
logging.error("`{0}` exists. `bwa samse` already run.".format(samfile))
return "", samfile
cmd = "bwa sampe " + " ".join((dbfile, sai1file, sai2file, \
read1file, read2file))
cmd += " " + opts.extra
if opts.cutoff:
cmd += " -a {0}".format(opts.cutoff)
if opts.uniq:
cmd += " -n 1"
return cmd, samfile
def mem(args, opts):
"""
%prog mem database.fasta read1.fq [read2.fq]
Wrapper for `bwa mem`. Output will be read1.sam.
"""
dbfile, read1file = args[:2]
readtype = opts.readtype
pl = readtype or "illumina"
pf = op.basename(read1file).split(".")[0]
rg = opts.rg or r"@RG\tID:{0}\tSM:sm\tLB:lb\tPL:{1}".format(pf, pl)
dbfile = check_index(dbfile)
args[0] = dbfile
samfile, _, unmapped = get_samfile(read1file,
dbfile,
bam=opts.bam,
unmapped=opts.unmapped)
if not need_update(read1file, samfile):
logging.error("`{0}` exists. `bwa mem` already run.".format(samfile))
return "", samfile
cmd = "{} mem".format(opts.bwa)
"""
-M Mark shorter split hits as secondary (for Picard compatibility).
"""
cmd += " -M -t {0}".format(opts.cpus)
cmd += ' -R "{0}"'.format(rg)
if readtype:
cmd += " -x {0}".format(readtype)
cmd += " " + opts.extra
cmd += " ".join(args)
return cmd, samfile
def mem(args, opts):
"""
%prog mem database.fasta read1.fq [read2.fq]
Wrapper for `bwa mem`. Output will be read1.sam.
"""
dbfile, read1file = args[:2]
pf = op.basename(read1file).split(".")[0]
rg = opts.rg or r"@RG\tID:{0}\tSM:sm\tLB:lb\tPL:illumina".format(pf)
dbfile = check_index(dbfile)
args[0] = dbfile
samfile, _, unmapped = get_samfile(read1file, dbfile,
bam=opts.bam, unmapped=opts.unmapped)
if not need_update(read1file, samfile):
logging.error("`{0}` exists. `bwa mem` already run.".format(samfile))
return "", samfile
cmd = "bwa mem " + " ".join(args)
cmd += " -M -t {0}".format(opts.cpus)
cmd += ' -R "{0}"'.format(rg)
if opts.readtype:
cmd += " -x {0}".format(opts.readtype)
cmd += " " + opts.extra
return cmd, samfile
def mem(args, opts):
"""
%prog mem database.fasta read1.fq [read2.fq]
Wrapper for `bwa mem`. Output will be read1.sam.
"""
dbfile, read1file = args[:2]
readtype = opts.readtype
pl = readtype or "illumina"
pf = op.basename(read1file).split(".")[0]
rg = opts.rg or r"@RG\tID:{0}\tSM:sm\tLB:lb\tPL:{1}".format(pf, pl)
dbfile = check_index(dbfile)
args[0] = dbfile
samfile, _, unmapped = get_samfile(read1file, dbfile,
bam=opts.bam, unmapped=opts.unmapped)
if not need_update(read1file, samfile):
logging.error("`{0}` exists. `bwa mem` already run.".format(samfile))
return "", samfile
cmd = "{} mem".format(opts.bwa)
'''
-M Mark shorter split hits as secondary (for Picard compatibility).
'''
cmd += " -M -t {0}".format(opts.cpus)
cmd += ' -R "{0}"'.format(rg)
if readtype:
cmd += " -x {0}".format(readtype)
cmd += " " + opts.extra
cmd += " ".join(args)
return cmd, samfile
def bwasw(args, opts):
"""
%prog bwasw database.fasta long_read.fastq
Wrapper for `bwa bwasw`. Output will be long_read.sam.
"""
dbfile, readfile = args
safile = check_index(dbfile)
samfile, _, unmapped = get_samfile(readfile, dbfile,
bam=opts.bam, unmapped=opts.unmapped)
if not need_update(safile, samfile):
logging.error("`{0}` exists. `bwa bwasw` already run.".format(samfile))
return "", samfile
cmd = "bwa bwasw " + " ".join(args)
cmd += " -t {0}".format(opts.cpus)
cmd += " " + opts.extra
return cmd, samfile
def mem(args, opts):
"""
%prog mem database.fasta read1.fq [read2.fq]
Wrapper for `bwa mem`. Output will be read1.sam.
"""
dbfile, read1file = args[:2]
check_index(dbfile)
samfile, _, unmapped = get_samfile(read1file, dbfile,
bam=opts.bam, unmapped=opts.unmapped)
if not need_update(read1file, samfile):
logging.error("`{0}` exists. `bwa mem` already run.".format(samfile))
return "", samfile
cmd = "bwa mem " + " ".join(args)
cmd += " -t {0}".format(opts.cpus)
cmd += " " + opts.extra
return cmd, samfile
def bwasw(args, opts):
"""
%prog bwasw database.fasta long_read.fastq
Wrapper for `bwa bwasw`. Output will be long_read.sam.
"""
dbfile, readfile = args
dbfile = check_index(dbfile)
samfile, _, unmapped = get_samfile(readfile, dbfile,
bam=opts.bam, unmapped=opts.unmapped)
if not need_update(dbfile, samfile):
logging.error("`{0}` exists. `bwa bwasw` already run.".format(samfile))
return "", samfile
cmd = "bwa bwasw " + " ".join(args)
cmd += " -t {0}".format(opts.cpus)
cmd += " " + opts.extra
return cmd, samfile
def samse(args, opts):
"""
%prog samse database.fasta short_read.fastq
Wrapper for `bwa samse`. Output will be short_read.sam.
"""
dbfile, readfile = args
safile = check_index(dbfile)
saifile = check_aln(dbfile, readfile, cpus=opts.cpus)
samfile, _, unmapped = get_samfile(readfile, dbfile,
bam=opts.bam, unmapped=opts.unmapped)
if not need_update((safile, saifile), samfile):
logging.error("`{0}` exists. `bwa samse` already run.".format(samfile))
return "", samfile
cmd = "bwa samse {0} {1} {2}".format(dbfile, saifile, readfile)
cmd += " " + opts.extra
if opts.uniq:
cmd += " -n 1"
return cmd, samfile
def samse(args, opts):
"""
%prog samse database.fasta short_read.fastq
Wrapper for `bwa samse`. Output will be short_read.sam.
"""
dbfile, readfile = args
dbfile = check_index(dbfile)
saifile = check_aln(dbfile, readfile, cpus=opts.cpus)
samfile, _, unmapped = get_samfile(readfile, dbfile,
bam=opts.bam, unmapped=opts.unmapped)
if not need_update((dbfile, saifile), samfile):
logging.error("`{0}` exists. `bwa samse` already run.".format(samfile))
return "", samfile
cmd = "bwa samse {0} {1} {2}".format(dbfile, saifile, readfile)
cmd += " " + opts.extra
if opts.uniq:
cmd += " -n 1"
return cmd, samfile
def mem(args, opts):
"""
%prog mem database.fasta read1.fq [read2.fq]
Wrapper for `bwa mem`. Output will be read1.sam.
"""
dbfile, read1file = args[:2]
check_index(dbfile)
samfile, _, unmapped = get_samfile(read1file,
dbfile,
bam=opts.bam,
unmapped=opts.unmapped)
if not need_update(read1file, samfile):
logging.error("`{0}` exists. `bwa mem` already run.".format(samfile))
return "", samfile
cmd = "bwa mem " + " ".join(args)
cmd += " -t {0}".format(opts.cpus)
if opts.readtype:
cmd += " -x {0}".format(opts.readtype)
cmd += " " + opts.extra
return cmd, samfile
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
