def gmap_setup(gsnap_dir, out_dir, ref_fasta):
ref_base = op.splitext(op.basename(ref_fasta))[0]
ref_dir = op.dirname(ref_fasta)
ref_name = op.basename(ref_base)
# have to cd to the out_dir because gsnap writes to cwd.
cmd = "set -e\n cd %(ref_dir)s && \n"
cmd += "gmap_build"
cmd += " -k 12 -D %(ref_dir)s -d %(ref_base)s %(ref_fasta)s > %(out_dir)s/gmap_build.log && "
cmd += "\ncmetindex -d %(ref_base)s -F %(ref_dir)s > gmap_cmetindex.log 2> gmap_cmetindex.error.log"
cmd %= locals()
print >>sys.stderr, "[ command ] $", cmd
cmd_last = op.join(out_dir, "ran_gsnap_setup.sh")
rerun = False
if not op.exists(cmd_last) or not is_up_to_date_b(ref_fasta, cmd_last) or not is_same_cmd(cmd, cmd_last):
fh = open(cmd_last, "w")
print >>fh, cmd
fh.close()
rerun = True
elif is_up_to_date_b(ref_fasta, cmd_last) and not is_same_cmd(cmd, cmd_last):
fh = open(cmd_last, "w")
print >>fh, cmd
fh.close()
rerun = True
# TODO: check time-stamp
rerun = True
if rerun:
p = Popen(cmd.replace('\n', ' '), shell=True)
print >>sys.stderr, "^ executing gmap/gsnap setup^"
if p.wait() != 0:
pass
else:
print >>sys.stderr, "gsnap setup stuff is up to date, re-using"
return ref_base
def gmap_setup(gsnap_dir, out_dir, ref_fasta):
ref_base = op.splitext(op.basename(ref_fasta))[0]
ref_dir = op.dirname(ref_fasta)
ref_name = op.basename(ref_base)
# have to cd to the out_dir because gsnap writes to cwd.
cmd = "set -e\n cd %(ref_dir)s && \n"
cmd += "gmap_build -w 2"
cmd += " -k 15 -D %(ref_dir)s -d %(ref_base)s %(ref_fasta)s > %(out_dir)s/gmap_build.log && "
cmd += "\ncmetindex -d %(ref_base)s -F %(ref_dir)s -k 15 > gmap_cmetindex.log 2> gmap_cmetindex.error.log"
cmd %= locals()
print >> sys.stderr, "[ command ] $", cmd
cmd_last = op.join(out_dir, "ran_gsnap_setup.sh")
rerun = False
if not op.exists(cmd_last) or not is_up_to_date_b(
ref_fasta, cmd_last) or not is_same_cmd(cmd, cmd_last):
fh = open(cmd_last, "w")
print >> fh, cmd
fh.close()
rerun = True
elif is_up_to_date_b(ref_fasta,
cmd_last) and not is_same_cmd(cmd, cmd_last):
fh = open(cmd_last, "w")
print >> fh, cmd
fh.close()
rerun = True
# TODO: check time-stamp
rerun = True
if rerun:
p = Popen(cmd.replace('\n', ' '), shell=True)
print >> sys.stderr, "^ executing gmap/gsnap setup^"
if p.wait() != 0:
pass
else:
print >> sys.stderr, "gsnap setup stuff is up to date, re-using"
return ref_base
def run_gsnap(gsnap_dir, gsnap_args, out_dir, ref_fasta, reads_paths,
cpu_count):
ref_base = op.splitext(ref_fasta)[0]
ref_name = op.basename(ref_base)
ref_dir = op.dirname(ref_fasta)
reads_paths = [op.abspath(r) for r in reads_paths]
log = op.join(out_dir, "gsnap_run.log")
if not "--mode" in gsnap_args:
cmet = "--mode=cmet-nonstranded"
print >>sys.stderr, "using non-stranded mode, specify " \
+ " --mode=cmet-stranded in --extra-args for stranded mode"
if any(r.endswith(".gz") for r in reads_paths):
gsnap_args += " --gunzip "
reads_paths_str = " ".join(reads_paths)
out_sam = op.abspath(op.join(out_dir, "methylcoded.gsnap.sam"))
cmd = "gsnap --quiet-if-excessive -A sam -k 15 "
cmd += " --nofails --nthreads %(cpu_count)i -D %(ref_dir)s %(gsnap_args)s"
cmd += " -d %(ref_name)s %(cmet)s %(reads_paths_str)s > %(out_sam)s 2> %(log)s"
cmd %= locals()
print >> sys.stderr, cmd
cmd_path = op.join(out_dir, "ran_gsnap.sh")
new_cmd = False
if not is_same_cmd(cmd, cmd_path):
fh = open(cmd_path, "w")
print >> fh, cmd
fh.close()
new_cmd = True
if not new_cmd and all(is_up_to_date_b(r, out_sam) for r in reads_paths) \
and all(is_up_to_date_b(r, cmd_path) for r in reads_paths):
print >> sys.stderr, "^ NOT executing gsnap. everything is up to date.^"
else:
fh = open(cmd_path, "w")
print >> fh, cmd
fh.close()
print >> sys.stderr, "^ executing gsnap. ^"
p = Popen(cmd.replace('\n', ' '), shell=True)
if p.wait() != 0:
os.unlink(out_sam)
print >> sys.stderr, "ERROR:\n", open(log).read()
sys.exit(1)
return out_sam
def run_gsnap(gsnap_dir, gsnap_args, out_dir, ref_fasta, reads_paths, cpu_count):
ref_base = op.splitext(ref_fasta)[0]
ref_name = op.basename(ref_base)
ref_dir = op.dirname(ref_fasta)
reads_paths = [op.abspath(r) for r in reads_paths]
log = op.join(out_dir, "gsnap_run.log")
if not "--mode" in gsnap_args:
cmet = "--mode=cmet-nonstranded"
print >>sys.stderr, "using non-stranded mode, specify " \
+ " --mode=cmet-stranded in --extra-args for stranded mode"
if any(r.endswith(".gz") for r in reads_paths):
gsnap_args += " --gunzip "
reads_paths_str = " ".join(reads_paths)
out_sam = op.abspath(op.join(out_dir, "methylcoded.gsnap.sam"))
cmd = "gsnap --quiet-if-excessive -A sam -k 15 "
cmd += " --nofails --nthreads %(cpu_count)i -D %(ref_dir)s %(gsnap_args)s"
cmd += " -d %(ref_name)s %(cmet)s %(reads_paths_str)s > %(out_sam)s 2> %(log)s"
cmd %= locals()
print >>sys.stderr, cmd
cmd_path = op.join(out_dir, "ran_gsnap.sh")
new_cmd = False
if not is_same_cmd(cmd, cmd_path):
fh = open(cmd_path, "w")
print >>fh, cmd
fh.close()
new_cmd = True
if not new_cmd and all(is_up_to_date_b(r, out_sam) for r in reads_paths) \
and all(is_up_to_date_b(r, cmd_path) for r in reads_paths):
print >>sys.stderr, "^ NOT executing gsnap. everything is up to date.^"
else:
fh = open(cmd_path, "w")
print >>fh, cmd
fh.close()
print >>sys.stderr, "^ executing gsnap. ^"
p = Popen(cmd.replace('\n', ' '), shell=True)
if p.wait() != 0:
os.unlink(out_sam)
print >>sys.stderr, "ERROR:\n", open(log).read()
sys.exit(1)
return out_sam
