def run_multiple(self, jobs, num_procs=1):
pid_list = []
for input_fastq, tophat_options, out_dir in jobs:
if len(jobs) == 1:
self.run_single(input_fastq,
([] if not tophat_options else tophat_options),
(self.out_dir if not out_dir else out_dir))
return
pid = _common.fork_and_run(
self.run_single, input_fastq,
([] if not tophat_options else tophat_options),
(self.out_dir if not out_dir else out_dir))
pid_list.append(pid)
_common.wait_for_slot(pid_list, num_procs)
_common.wait_for_slot(pid_list, 0, True)
def run_multiple(self, jobs, num_procs = 1):
pid_list = []
for input_fastq, tophat_options, out_dir in jobs:
if len(jobs) == 1:
self.run_single(input_fastq,
([] if not tophat_options else tophat_options),
(self.out_dir if not out_dir else out_dir))
return
pid = _common.fork_and_run(self.run_single,
input_fastq,
([] if not tophat_options else tophat_options),
(self.out_dir if not out_dir else out_dir))
pid_list.append(pid)
_common.wait_for_slot(pid_list, num_procs)
_common.wait_for_slot(pid_list, 0, True)
def all_chrom(self, input_bam, scripture_options=[]):
out_file = os.path.join(self.out_dir, "transcripts.bed")
pid_list = []
for chrom in self.chroms:
pid = _common.fork_and_run(self.single_chrom, input_bam, chrom,
scripture_options)
pid_list.append(pid)
_common.wait_for_slot(pid_list, self.num_procs)
_common.wait_for_slot(pid_list, 0, True)
with open(out_file, "w") as final_fp:
for chrom in self.chroms:
in_file = os.path.join(self.out_dir, chrom + ".segments")
if not os.path.exists(in_file):
continue
final_fp.write(open(in_file).read())
def all_chrom(self, input_bam, scripture_options = []):
out_file = os.path.join(self.out_dir, "transcripts.bed")
pid_list = []
for chrom in self.chroms:
pid = _common.fork_and_run(self.single_chrom,
input_bam,
chrom,
scripture_options)
pid_list.append(pid)
_common.wait_for_slot(pid_list, self.num_procs)
_common.wait_for_slot(pid_list, 0, True)
with open(out_file, "w") as final_fp:
for chrom in self.chroms:
in_file = os.path.join(self.out_dir, chrom + ".segments")
if not os.path.exists(in_file):
continue
final_fp.write(open(in_file).read())
def main():
parse_options(sys.argv[1:])
if not os.path.exists(options.output_dir):
os.mkdir(options.output_dir)
# tophat runs
if not options.skip_tophat:
jobs = []
for label, run_params in zip(options.labels, options.runs):
run_out_dir = os.path.join(options.output_dir, label)
run_input_fastq = [run_params["left_reads"]]
extra_params = ["--seed-length", run_params["seed_len"]]
if run_params["right_reads"] != "None":
run_input_fastq.append(run_params["right_reads"])
extra_params.extend(["--mate-inner-dist", run_params["inner_dist"],
"--mate-std-dev", run_params["inner_dist_sd"]])
jobs.append((run_input_fastq,
["--GTF", options.reference,
"--no-novel-juncs",
"--no-novel-indels",
"--no-coverage-search"] + \
extra_params,
run_out_dir))
th = tophat.TopHat(options.bowtie_index,
options.output_dir,
options.num_threads)
th.run_multiple(jobs, options.num_procs)
# cufflinks runs
if not options.skip_cufflinks:
for label, run_params in zip(options.labels, options.runs):
input_bam = os.path.join(options.output_dir, label, "accepted_hits.bam")
if not os.path.exists(input_bam):
print "Error: Could not find %s" % (input_bam,)
continue
run_out_dir = os.path.join(options.output_dir, "FPKM_" + label)
cl = cufflinks.Cufflinks(run_out_dir, options.num_threads)
cl.run(input_bam, ["--min-frags-per-transfrag", "0",
"-G", options.reference] + \
([] if not options.mask else ["-M", options.mask]))
# make TDF files
if not options.skip_tdf:
# collect sample data
sample_info = []
for label, run_params in zip(options.labels, options.runs):
input_bam = os.path.join(options.output_dir, label, "accepted_hits.bam")
if not os.path.exists(input_bam):
print "Error: Could not find %s" % (input_bam,)
continue
wig_file = tempfile.NamedTemporaryFile("w", suffix=".wig", delete=False)
tdf_out = os.path.join(options.output_dir, label, label + ".tdf")
sample_info.append((input_bam, wig_file, tdf_out))
# make wigs
st = samtools.SAMTools()
pid_list = []
for input_bam, wig_file, tdf_out in sample_info:
pid = _common.fork_and_run(st.wig,
input_bam,
wig_file,
options.filter_less_than)
pid_list.append(pid)
_common.wait_for_slot(pid_list, options.num_threads)
_common.wait_for_slot(pid_list, 0, True)
# make tdfs
igv = igvtools.IGVTools()
pid_list = []
for input_bam, wig_file, tdf_out in sample_info:
pid = _common.fork_and_run(igv.tile,
wig_file.name,
tdf_out,
args[0])
pid_list.append(pid)
_common.wait_for_slot(pid_list, options.num_threads)
_common.wait_for_slot(pid_list, 0, True)
# remove tempfiles
for _, wig_file, _ in sample_info:
os.unlink(wig_file.name)
def main():
parse_options(sys.argv[1:])
if not os.path.exists(options.output_dir):
os.mkdir(options.output_dir)
# tophat runs
if not options.skip_tophat:
jobs = []
for label, run_params in zip(options.labels, options.runs):
run_out_dir = os.path.join(options.output_dir, label)
run_input_fastq = [run_params["left_reads"]]
extra_params = ["--seed-length", run_params["seed_len"]]
if run_params["right_reads"] != "None":
run_input_fastq.append(run_params["right_reads"])
extra_params.extend([
"--mate-inner-dist", run_params["inner_dist"],
"--mate-std-dev", run_params["inner_dist_sd"]
])
jobs.append((run_input_fastq,
["--GTF", options.reference,
"--no-novel-juncs",
"--no-novel-indels",
"--no-coverage-search"] + \
extra_params,
run_out_dir))
th = tophat.TopHat(options.bowtie_index, options.output_dir,
options.num_threads)
th.run_multiple(jobs, options.num_procs)
# cufflinks runs
if not options.skip_cufflinks:
for label, run_params in zip(options.labels, options.runs):
input_bam = os.path.join(options.output_dir, label,
"accepted_hits.bam")
if not os.path.exists(input_bam):
print "Error: Could not find %s" % (input_bam, )
continue
run_out_dir = os.path.join(options.output_dir, "FPKM_" + label)
cl = cufflinks.Cufflinks(run_out_dir, options.num_threads)
cl.run(input_bam, ["--min-frags-per-transfrag", "0",
"-G", options.reference] + \
([] if not options.mask else ["-M", options.mask]))
# make TDF files
if not options.skip_tdf:
# collect sample data
sample_info = []
for label, run_params in zip(options.labels, options.runs):
input_bam = os.path.join(options.output_dir, label,
"accepted_hits.bam")
if not os.path.exists(input_bam):
print "Error: Could not find %s" % (input_bam, )
continue
wig_file = tempfile.NamedTemporaryFile("w",
suffix=".wig",
delete=False)
tdf_out = os.path.join(options.output_dir, label, label + ".tdf")
sample_info.append((input_bam, wig_file, tdf_out))
# make wigs
st = samtools.SAMTools()
pid_list = []
for input_bam, wig_file, tdf_out in sample_info:
pid = _common.fork_and_run(st.wig, input_bam, wig_file,
options.filter_less_than)
pid_list.append(pid)
_common.wait_for_slot(pid_list, options.num_threads)
_common.wait_for_slot(pid_list, 0, True)
# make tdfs
igv = igvtools.IGVTools()
pid_list = []
for input_bam, wig_file, tdf_out in sample_info:
pid = _common.fork_and_run(igv.tile, wig_file.name, tdf_out,
args[0])
pid_list.append(pid)
_common.wait_for_slot(pid_list, options.num_threads)
_common.wait_for_slot(pid_list, 0, True)
# remove tempfiles
for _, wig_file, _ in sample_info:
os.unlink(wig_file.name)