def __init__(self,
reads,
outdir,
threads=1,
spades_kmers=None,
careful=True,
only_assembler=True,
verbose=False,
spades_use_first_success=False,
assembler='spades',
genomeSize=100000, # only matters for Canu if correcting reads (which we're not)
data_type='pacbio-corrected',
):
self.outdir = os.path.abspath(outdir)
self.reads = os.path.abspath(reads)
if not os.path.exists(self.reads):
raise Error('Reads file not found:' + self.reads)
self.verbose = verbose
self.samtools = external_progs.make_and_check_prog('samtools', verbose=self.verbose)
self.threads = threads
self.assembler = assembler
if self.assembler == 'spades':
self.spades = external_progs.make_and_check_prog('spades', verbose=self.verbose, required=True)
self.spades_kmers = self._build_spades_kmers(spades_kmers)
self.spades_use_first_success = spades_use_first_success
self.careful = careful
self.only_assembler = only_assembler
elif self.assembler == 'canu':
self.canu = external_progs.make_and_check_prog('canu', verbose=self.verbose, required=True)
self.genomeSize=genomeSize
self.data_type = data_type
else:
raise Error('Unknown assembler: "' + self.assembler + '". cannot continue')
def bwa_mem(
ref,
reads,
outfile,
threads=1,
bwa_options = '-x pacbio',
verbose=False,
index=None
):
samtools = external_progs.make_and_check_prog('samtools', verbose=verbose)
bwa = external_progs.make_and_check_prog('bwa', verbose=verbose)
unsorted_bam = outfile + '.tmp.unsorted.bam'
tmp_index = outfile + '.tmp.bwa_index'
bwa_index(ref, outprefix=tmp_index, verbose=verbose, bwa=bwa.exe())
cmd = ' '.join([
bwa.exe(), 'mem',
bwa_options,
'-t', str(threads),
tmp_index,
reads,
'|',
samtools.exe(), 'view',
'-F 0x0800',
'-T', ref,
'-b',
'-o', unsorted_bam,
'-',
])
common.syscall(cmd, verbose=verbose)
bwa_index_clean(tmp_index)
threads = min(4, threads)
thread_mem = int(500 / threads)
cmd = ' '.join([
samtools.exe(), 'sort',
'-@', str(threads),
'-m', str(thread_mem) + 'M',
unsorted_bam,
outfile[:-4]
])
common.syscall(cmd, verbose=verbose)
os.unlink(unsorted_bam)
cmd = samtools.exe() + ' index ' + outfile
common.syscall(cmd, verbose=verbose)
def __init__(
self,
reads,
outdir,
threads=1,
spades_kmers=None,
careful=True,
only_assembler=True,
verbose=False,
spades_use_first_success=False,
assembler='spades',
genomeSize=100000, # only matters for Canu if correcting reads (which we're not)
data_type='pacbio-raw',
):
self.outdir = os.path.abspath(outdir)
self.reads = os.path.abspath(reads)
if not os.path.exists(self.reads):
raise Error('Reads file not found:' + self.reads)
self.verbose = verbose
self.samtools = external_progs.make_and_check_prog(
'samtools', verbose=self.verbose)
self.threads = threads
self.assembler = assembler
if self.assembler == 'spades':
self.spades = external_progs.make_and_check_prog(
'spades', verbose=self.verbose, required=True)
self.spades_kmers = self._build_spades_kmers(spades_kmers)
self.spades_use_first_success = spades_use_first_success
self.careful = careful
self.only_assembler = only_assembler
elif self.assembler == 'canu':
self.canu = external_progs.make_and_check_prog(
'canu', verbose=self.verbose, required=True)
self.genomeSize = genomeSize
self.data_type = data_type
elif self.assembler == 'racon':
self.racon = external_progs.make_and_check_prog(
'racon', verbose=self.verbose, required=True)
self.data_type = data_type
else:
raise Error('Unknown assembler: "' + self.assembler +
'". cannot continue')
def __init__(
self,
reads,
outdir,
threads=1,
CanuError=0.045,
spades_kmers=None,
careful=True,
only_assembler=True,
verbose=False,
spades_use_first_success=False,
useCanu=False,
genomeSize=100000,
dataType='pacbio-raw',
):
self.outdir = os.path.abspath(outdir)
self.reads = os.path.abspath(reads)
if not os.path.exists(self.reads):
raise Error('Reads file not found:' + self.reads)
self.verbose = verbose
self.samtools = external_progs.make_and_check_prog(
'samtools', verbose=self.verbose)
if (not useCanu):
self.useCanu = False
self.assembler = 'spades'
self.spades = external_progs.make_and_check_prog(
'spades', verbose=self.verbose)
self.spades_kmers = self._build_spades_kmers(spades_kmers)
self.spades_use_first_success = spades_use_first_success
self.careful = careful
self.only_assembler = only_assembler
self.threads = threads
else:
self.CanuError = CanuError
self.useCanu = True
self.assembler = 'canu'
self.canu = external_progs.make_and_check_prog(
'canu', verbose=self.verbose)
self.genomeSize = genomeSize
#self.genomeSize=self.length_cutoff
self.dataType = dataType
def __init__(self,
reads,
outdir,
threads=1,
spades_kmers=None,
verbose=False,
spades_use_first_success=False,
):
self.outdir = os.path.abspath(outdir)
self.reads = os.path.abspath(reads)
if not os.path.exists(self.reads):
raise Error('Reads file not found:' + self.reads)
self.verbose = verbose
self.threads = threads
self.spades = external_progs.make_and_check_prog('spades', verbose=self.verbose)
self.spades_kmers = self._build_spades_kmers(spades_kmers)
self.spades_use_first_success = spades_use_first_success
self.samtools = external_progs.make_and_check_prog('samtools', verbose=self.verbose)
self.assembler = 'spades'
def __init__(
self,
reads,
outdir,
threads=1,
spades_kmers=None,
verbose=False,
spades_use_first_success=False,
):
self.outdir = os.path.abspath(outdir)
self.reads = os.path.abspath(reads)
if not os.path.exists(self.reads):
raise Error('Reads file not found:' + self.reads)
self.verbose = verbose
self.threads = threads
self.spades = external_progs.make_and_check_prog('spades',
verbose=self.verbose)
self.spades_kmers = self._build_spades_kmers(spades_kmers)
self.spades_use_first_success = spades_use_first_success
self.samtools = external_progs.make_and_check_prog(
'samtools', verbose=self.verbose)
self.assembler = 'spades'
def bwa_index(infile, outprefix=None, bwa=None, verbose=False):
if bwa is None:
bwa = external_progs.make_and_check_prog('bwa', verbose=verbose)
if outprefix is None:
outprefix = infile
missing = [
not os.path.exists(outprefix + '.' + x) for x in index_extensions
]
if True not in missing:
return
cmd = ' '.join([bwa.exe(), 'index', '-p', outprefix, infile])
common.syscall(cmd, verbose=verbose)
def bwa_index(infile, outprefix=None, bwa=None, verbose=False):
if bwa is None:
bwa = external_progs.make_and_check_prog('bwa', verbose=verbose)
if outprefix is None:
outprefix = infile
missing = [not os.path.exists(outprefix + '.' + x) for x in index_extensions]
if True not in missing:
return
cmd = ' '.join([
bwa.exe(), 'index',
'-p', outprefix,
infile
])
common.syscall(cmd, verbose=verbose)
def bwa_mem(
ref,
reads,
outfile,
threads=1,
bwa_options = '-x pacbio',
verbose=False,
index=None
):
samtools = external_progs.make_and_check_prog('samtools', verbose=verbose)
bwa = external_progs.make_and_check_prog('bwa', verbose=verbose)
unsorted_bam = outfile + '.tmp.unsorted.bam'
tmp_index = outfile + '.tmp.bwa_index'
bwa_index(ref, outprefix=tmp_index, verbose=verbose, bwa=bwa)
cmd = ' '.join([
bwa.exe(), 'mem',
bwa_options,
'-t', str(threads),
tmp_index,
reads,
'|',
samtools.exe(), 'view',
'-F 0x0800',
'-T', ref,
'-b',
'-o', unsorted_bam,
'-',
])
common.syscall(cmd, verbose=verbose)
bwa_index_clean(tmp_index)
threads = min(4, threads)
thread_mem = int(500 / threads)
# here we have to check for the version of samtools, starting from 1.3 the
# -o flag is used for specifying the samtools sort output-file.
# Starting from 1.2 you can use the -o flag, but can't have
# -o out.bam at the end of the call, so use new style from 1.3 onwards.
outparam = ''
if samtools.version_at_least('1.3'):
outparam = '-o'
samout = outfile
else:
samout = outfile[:-4]
cmd = ' '.join([
samtools.exe(), 'sort',
'-@', str(threads),
'-m', str(thread_mem) + 'M',
unsorted_bam,
outparam,samout
])
common.syscall(cmd, verbose=verbose)
os.unlink(unsorted_bam)
cmd = samtools.exe() + ' index ' + outfile
common.syscall(cmd, verbose=verbose)
def bwa_mem(ref,
reads,
outfile,
threads=1,
bwa_options='-x pacbio',
verbose=False,
index=None):
samtools = external_progs.make_and_check_prog('samtools', verbose=verbose)
bwa = external_progs.make_and_check_prog('bwa', verbose=verbose)
unsorted_bam = outfile + '.tmp.unsorted.bam'
tmp_index = outfile + '.tmp.bwa_index'
bwa_index(ref, outprefix=tmp_index, verbose=verbose, bwa=bwa)
cmd = ' '.join([
bwa.exe(),
'mem',
bwa_options,
'-t',
str(threads),
tmp_index,
reads,
'|',
samtools.exe(),
'view',
'-F 0x0800',
'-T',
ref,
'-b',
'-o',
unsorted_bam,
'-',
])
common.syscall(cmd, verbose=verbose)
bwa_index_clean(tmp_index)
threads = min(4, threads)
thread_mem = int(500 / threads)
# here we have to check for the version of samtools, starting from 1.3 the
# -o flag is used for specifying the samtools sort output-file.
# Starting from 1.2 you can use the -o flag, but can't have
# -o out.bam at the end of the call, so use new style from 1.3 onwards.
outparam = ''
if samtools.version_at_least('1.3'):
outparam = '-o'
samout = outfile
else:
samout = outfile[:-4]
cmd = ' '.join([
samtools.exe(), 'sort', '-@',
str(threads), '-m',
str(thread_mem) + 'M', unsorted_bam, outparam, samout
])
common.syscall(cmd, verbose=verbose)
os.unlink(unsorted_bam)
cmd = samtools.exe() + ' index ' + outfile
common.syscall(cmd, verbose=verbose)