def convert_fasta(ref_fasta, just_name=False):
out_fa = ref_fasta + ".bwameth.c2t"
if just_name:
return out_fa
msg = "c2t in %s to %s" % (ref_fasta, out_fa)
if is_newer_b(ref_fasta, out_fa):
sys.stderr.write("already converted: %s\n" % msg)
return out_fa
sys.stderr.write("converting %s\n" % msg)
try:
fh = open(out_fa, "w")
for header, seq in fasta_iter(ref_fasta):
########### Reverse ######################
fh.write(">r%s\n" % header)
#if non_cpg_only:
# for ctx in "TAG": # use "ATC" for fwd
# seq = seq.replace('G' + ctx, "A" + ctx)
# for line in wrap(seq):
# print >>fh, line
#else:
for line in wrap(seq.replace("G", "A")):
fh.write(line + '\n')
########### Forward ######################
fh.write(">f%s\n" % header)
for line in wrap(seq.replace("C", "T")):
fh.write(line + '\n')
fh.close()
except:
fh.close(); os.unlink(out_fa)
raise
return out_fa
def bwa_mem(fa,
fq_convert_cmd,
extra_args,
threads=1,
rg=None,
paired=True,
set_as_failed=None,
do_not_penalize_chimeras=False):
conv_fa = convert_fasta(fa, just_name=True)
if not is_newer_b(conv_fa, (conv_fa + '.amb', conv_fa + '.sa')):
raise BWAMethException("first run bwameth.py index %s" % fa)
if not rg is None and not rg.startswith('@RG'):
rg = '@RG\\tID:{rg}\\tSM:{rg}'.format(rg=rg)
#starts the pipeline with the program to convert fastqs
cmd = ("|%s " % fq_convert_cmd)
# penalize clipping and unpaired. lower penalty on mismatches (-B)
cmd += "|bwa mem -T 40 -B 2 -L 10 -CM "
if paired:
cmd += ("-U 100 -p ")
cmd += "-R '{rg}' -t {threads} {extra_args} {conv_fa} -"
cmd = cmd.format(**locals())
sys.stderr.write("running: %s\n" % cmd.lstrip("|"))
as_bam(cmd, fa, set_as_failed, do_not_penalize_chimeras)
def bwa_mem(fa,
mfq,
extra_args,
prefix='bwa-meth',
threads=1,
rg=None,
calmd=False,
paired=True,
set_as_failed=None):
conv_fa = convert_fasta(fa, just_name=True)
if not is_newer_b(conv_fa, (conv_fa + '.amb', conv_fa + '.sa')):
raise BWAMethException("first run bwameth.py index %s" % fa)
if not rg is None and not rg.startswith('@RG'):
rg = '@RG\tID:{rg}\tSM:{rg}'.format(rg=rg)
# penalize clipping and unpaired. lower penalty on mismatches (-B)
cmd = "|bwa mem -T 40 -B 2 -L 10 -CM "
if paired:
cmd += ("-U 100 -p ")
cmd += "-R '{rg}' -t {threads} {extra_args} {conv_fa} {mfq}"
cmd = cmd.format(**locals())
sys.stderr.write("running: %s\n" % cmd.lstrip("|"))
as_bam(cmd, fa, prefix, calmd, set_as_failed)
def bwa_index(fa):
if is_newer_b(fa, (fa + '.amb', fa + '.sa')):
return
sys.stderr.write("indexing: %s\n" % fa)
try:
run("bwa index -a bwtsw %s" % fa)
except:
if op.exists(fa + ".amb"):
os.unlink(fa + ".amb")
raise
def bwa_index(fa):
if is_newer_b(fa, (fa + '.amb', fa + '.sa')):
return
sys.stderr.write("indexing: %s\n" % fa)
try:
run("bwa index -a bwtsw %s" % fa)
except:
if op.exists(fa + ".amb"):
os.unlink(fa + ".amb")
raise
def bwa_mem(fa, mfq, extra_args, prefix='bwa-meth', threads=1, rg=None,
calmd=False, paired=True, set_as_failed=None):
conv_fa = convert_fasta(fa, just_name=True)
if not is_newer_b(conv_fa, (conv_fa + '.amb', conv_fa + '.sa')):
raise BWAMethException("first run bwameth.py index %s" % fa)
# penalize clipping and unpaired. lower penalty on mismatches (-B)
cmd = "|bwa mem -T 40 -B 2 -L 10 -CM "
if paired:
cmd += ("-U 100 -p ")
cmd += "-t {threads} {extra_args} {conv_fa} {mfq}"
cmd = cmd.format(**locals())
sys.stderr.write("running: %s\n" % cmd.lstrip("|"))
as_bam(cmd, fa, prefix, calmd, set_as_failed)
def bwa_mem(fa, mfq, extra_args, threads=1, rg=None,
paired=True, set_as_failed=None):
conv_fa = convert_fasta(fa, just_name=True)
if not is_newer_b(conv_fa, (conv_fa + '.amb', conv_fa + '.sa')):
raise BWAMethException("first run bwameth.py index %s" % fa)
if not rg is None and not rg.startswith('@RG'):
rg = '@RG\tID:{rg}\tSM:{rg}'.format(rg=rg)
# penalize clipping and unpaired. lower penalty on mismatches (-B)
cmd = "|bwa mem -T 40 -B 2 -L 10 -CM "
if paired:
cmd += ("-U 100 -p ")
cmd += "-R '{rg}' -t {threads} {extra_args} {conv_fa} {mfq}"
cmd = cmd.format(**locals())
sys.stderr.write("running: %s\n" % cmd.lstrip("|"))
as_bam(cmd, fa, set_as_failed)
def convert_fasta(ref_fasta, just_name=False):
out_fa = ref_fasta + ".bwameth.c2t"
if just_name:
return out_fa
msg = "c2t in %s to %s" % (ref_fasta, out_fa)
if is_newer_b(ref_fasta, out_fa):
sys.stderr.write("already converted: %s\n" % msg)
return out_fa
sys.stderr.write("converting %s\n" % msg)
try:
fh = open(out_fa, "w")
for header, seq in fasta_iter(ref_fasta):
########### Reverse ######################
fh.write(">r%s\n" % header)
#if non_cpg_only:
# for ctx in "TAG": # use "ATC" for fwd
# seq = seq.replace('G' + ctx, "A" + ctx)
# for line in wrap(seq):
# print >>fh, line
#else:
for line in wrap(seq.replace("G", "A")):
fh.write(line + '\n')
########### Forward ######################
fh.write(">f%s\n" % header)
for line in wrap(seq.replace("C", "T")):
fh.write(line + '\n')
fh.close()
except:
try:
fh.close()
except UnboundLocalError:
pass
os.unlink(out_fa)
raise
return out_fa
def bwa_mem(fa,
fq_convert_cmd,
extra_args,
threads=1,
minSeedLength=19,
bandWidth=100,
dropoff=100,
mismatchPenalty=2,
gapOpenPenalty=6,
gapExtensionPenalty=1,
clippingPenalty=10,
minAlignmentScore=40,
rg=None,
paired=True,
set_as_failed=None):
conv_fa = convert_fasta(fa, just_name=True)
if not is_newer_b(conv_fa, (conv_fa + '.amb', conv_fa + '.sa')):
raise BWAMethException("first run bwameth.py index %s" % fa)
if not rg is None and not rg.startswith('@RG'):
rg = '@RG\\tID:{rg}\\tSM:{rg}'.format(rg=rg)
#starts the pipeline with the program to convert fastqs
cmd = ("|%s " % fq_convert_cmd)
# penalize clipping and unpaired. lower penalty on mismatches (-B)
cmd += "|bwa mem -CM "
if paired:
cmd += ("-U 100 -p ")
cmd += f"-R '{rg}' -t {threads} -k {minSeedLength} -w {bandWidth} -d {dropoff} "
cmd += f"-B {mismatchPenalty} -O {gapOpenPenalty} -E {gapExtensionPenalty} "
cmd += f"-L {clippingPenalty} -T {minAlignmentScore} {extra_args} {conv_fa} -"
#cmd = cmd.format(**locals())
sys.stderr.write("running: %s\n" % cmd.lstrip("|"))
as_bam(cmd, fa, set_as_failed)