def trim_primers_and_adapters(fasta_in,
fasta_out,
adapters_fa,
primers_fa,
min_length=100,
min_dist_to_end=25,
window_length=10,
min_pc=90):
'''Trim adapers and/or primers off contig ends'''
assert adapters_fa is not None or primers_fa is not None
tmpdir = tempfile.mkdtemp(prefix='tmp.trim.', dir=os.getcwd())
tmp_prefix = os.path.join(tmpdir, 'out')
if adapters_fa is None:
trim_query = primers_fa
elif primers_fa is None:
trim_query = adapters_fa
else:
trim_query = tmp_prefix + '.query.fa'
common.syscall('cat ' + adapters_fa + ' ' + primers_fa + ' > ' +
trim_query)
_trim_ends(fasta_in,
fasta_out,
trim_query,
min_length=min_length,
min_dist_to_end=min_dist_to_end,
window_length=window_length,
min_pc=min_pc)
shutil.rmtree(tmpdir)
def trim_primers_and_adapters(fasta_in, fasta_out, adapters_fa, primers_fa, min_length=100, min_dist_to_end=25, window_length=10, min_pc=90):
'''Trim adapers and/or primers off contig ends'''
assert adapters_fa is not None or primers_fa is not None
tmpdir = tempfile.mkdtemp(prefix='tmp.trim.', dir=os.getcwd())
tmp_prefix = os.path.join(tmpdir, 'out')
if adapters_fa is None:
trim_query = primers_fa
elif primers_fa is None:
trim_query = adapters_fa
else:
trim_query = tmp_prefix + '.query.fa'
common.syscall('cat ' + adapters_fa + ' ' + primers_fa + ' > ' + trim_query)
_trim_ends(fasta_in, fasta_out, trim_query, min_length=min_length, min_dist_to_end=min_dist_to_end, window_length=window_length, min_pc=min_pc)
shutil.rmtree(tmpdir)
def _run_kmc_with_script(script,
reads,
outfile,
kmer,
min_count,
max_count,
m_option,
verbose,
allow_fail,
threads=1):
f = pyfastaq.utils.open_file_write(script)
print('set -e', file=f)
kmc_command = ''.join([
'kmc -fa', ' -m',
str(m_option), ' -k',
str(kmer), ' -sf',
str(threads), ' -ci',
str(min_count), ' -cs',
str(max_count), ' -cx',
str(max_count), ' ', reads, ' kmc_out', ' $PWD'
])
print('ulimit -n 4096', file=f)
print(kmc_command, end='', file=f)
if verbose >= 2:
print('', file=f)
print('run kmc:', os.getcwd(), kmc_command)
else:
print(' > /dev/null', file=f)
print('kmc_dump', 'kmc_out', 'kmc_out.dump', file=f)
print('sort -k2nr', 'kmc_out.dump >', outfile, file=f)
pyfastaq.utils.close(f)
return common.syscall('bash ' + script, allow_fail=allow_fail)
def _run_kmc_with_script(script, reads, outfile, kmer, min_count, max_count, m_option, verbose, allow_fail):
f = pyfastaq.utils.open_file_write(script)
print('set -e', file=f)
kmc_command = ''.join([
'kmc -fa',
' -m', str(m_option),
' -k', str(kmer),
' -sf', '1',
' -ci', str(min_count),
' -cs', str(max_count),
' -cx', str(max_count),
' ', reads,
' kmc_out',
' $PWD'
])
print(kmc_command, end='', file=f)
if verbose >= 2:
print('', file=f)
print('run kmc:', os.getcwd(), kmc_command)
else:
print(' > /dev/null', file=f)
print('kmc_dump', 'kmc_out', 'kmc_out.dump', file=f)
print('sort -k2nr', 'kmc_out.dump >', outfile, file=f)
pyfastaq.utils.close(f)
return common.syscall('bash ' + script, allow_fail=allow_fail)
def run_nucmer(query, ref, outfile, min_id=95, min_length=100, breaklen=200):
query = os.path.abspath(query)
ref = os.path.abspath(ref)
outfile = os.path.abspath(outfile)
tmpdir = tempfile.mkdtemp(prefix='tmp.run_nucmer.', dir=os.getcwd())
original_dir = os.getcwd()
os.chdir(tmpdir)
script = 'run_nucmer.sh'
f = pyfastaq.utils.open_file_write(script)
print('nucmer --maxmatch -p p -b', breaklen, ref, query, file=f)
print('delta-filter -i', min_id, '-l', min_length, 'p.delta > p.delta.filter', file=f)
print('show-coords -dTlro p.delta.filter >', outfile, file=f)
pyfastaq.utils.close(f)
common.syscall('bash ' + script)
os.chdir(original_dir)
shutil.rmtree(tmpdir)
def run_nucmer(query, ref, outfile, min_id=95, min_length=100, breaklen=200):
query = os.path.abspath(query)
ref = os.path.abspath(ref)
outfile = os.path.abspath(outfile)
tmpdir = tempfile.mkdtemp(prefix='tmp.run_nucmer.', dir=os.getcwd())
original_dir = os.getcwd()
os.chdir(tmpdir)
script = 'run_nucmer.sh'
f = pyfastaq.utils.open_file_write(script)
print('nucmer --maxmatch -p p -b', breaklen, ref, query, file=f)
print('delta-filter -i', min_id, '-l', min_length, 'p.delta > p.delta.filter', file=f)
print('show-coords -dTlro p.delta.filter >', outfile, file=f)
pyfastaq.utils.close(f)
common.syscall('bash ' + script)
os.chdir(original_dir)
shutil.rmtree(tmpdir)
def run_trimmomatic(reads1, reads2, outprefix, trimmo_jar, adapters, minlen=50, verbose=0, threads=1, qual_trim=''):
cmd = ' '.join([
'java -Xmx1000m -jar',
trimmo_jar,
'PE',
'-threads', str(threads),
reads1,
reads2,
outprefix + '_1.fq',
outprefix + '.unpaired_1.fq',
outprefix + '_2.fq',
outprefix + '.unpaired_2.fq',
'ILLUMINACLIP:' + os.path.abspath(adapters) + ':2:10:7:1',
qual_trim,
'MINLEN:' + str(minlen)
])
if verbose:
print('Run trimmomatic:', cmd)
common.syscall(cmd)
os.unlink(outprefix + '.unpaired_1.fq')
os.unlink(outprefix + '.unpaired_2.fq')
def run_trimmomatic(reads1,
reads2,
outprefix,
trimmo_jar,
adapters,
minlen=50,
verbose=0,
threads=1,
qual_trim=''):
cmd = ' '.join([
'java -Xmx1000m -jar', trimmo_jar, 'PE', '-threads',
str(threads), reads1, reads2, outprefix + '_1.fq',
outprefix + '.unpaired_1.fq', outprefix + '_2.fq',
outprefix + '.unpaired_2.fq',
'ILLUMINACLIP:' + os.path.abspath(adapters) + ':2:10:7:1', qual_trim,
'MINLEN:' + str(minlen)
])
if verbose:
print('Run trimmomatic:', cmd)
common.syscall(cmd)
os.unlink(outprefix + '.unpaired_1.fq')
os.unlink(outprefix + '.unpaired_2.fq')
def map_reads(reads_fwd, reads_rev, ref_fa, out_prefix, index_k=15, index_s=3, threads=1, max_insert=1000, minid=0.5, verbose=0, required_flag=None, sort=False, exclude_flag=None, mate_ref=None, extra_smalt_map_ops=None):
if extra_smalt_map_ops is None:
extra_smalt_map_ops = ''
map_index = out_prefix + '.map_index'
clean_files = [map_index + '.' + x for x in ['smi', 'sma']]
index_cmd = ' '.join([
'smalt index',
'-k', str(index_k),
'-s', str(index_s),
map_index,
ref_fa
])
map_cmd = 'smalt map ' + extra_smalt_map_ops + ' '
# depending on OS, -n can break smalt, so only use -n if it's > 1.
if threads > 1:
map_cmd += '-n ' + str(threads) + ' -O '
if reads_rev is None:
map_cmd += ' '.join([
'-y', str(minid),
map_index,
reads_fwd,
])
else:
map_cmd += ' '.join([
'-i', str(max_insert),
'-y', str(minid),
map_index,
reads_fwd,
reads_rev,
])
if mate_ref is not None:
map_cmd += r''' | awk '$7=="''' + mate_ref + '"\''
map_cmd += ' | samtools view'
if required_flag is not None:
map_cmd += ' -f ' + str(required_flag)
if exclude_flag is not None:
map_cmd += ' -F ' + str(exclude_flag)
final_bam = out_prefix + '.bam'
if sort:
intermediate_bam = out_prefix + '.unsorted.bam'
else:
intermediate_bam = final_bam
map_cmd += ' -bS -T ' + ref_fa + ' - > ' + intermediate_bam
common.syscall(index_cmd)
common.syscall(map_cmd)
if verbose >= 2:
print(' map reads. Index: ', index_cmd)
print(' map reads. Mapping:', map_cmd)
if sort:
threads = min(4, threads)
thread_mem = int(500 / threads)
sort_cmd = 'samtools sort -@' + str(threads) + ' -m ' + str(thread_mem) + 'M ' + intermediate_bam + ' ' + out_prefix
index_cmd = 'samtools index ' + final_bam
if verbose >= 2:
print(' map reads. sort: ', sort_cmd)
common.syscall(sort_cmd)
if verbose >= 2:
print(' map reads. index: ', index_cmd)
common.syscall(index_cmd)
for fname in clean_files:
os.unlink(fname)