def run(self, debug=False):
pool = Pool(
processes=cpu(use_mem=3221225472, cpu_limit=len(self.contigs)))
for chrom in self.contigs:
pool.apply_async(self.win_correct, args=(chrom, ))
pool.close()
pool.join()
for chrom in self.contigs:
cnvdata = dict()
nbarg = os.path.join(self.indir, "%s.nbinom.arg" % chrom)
if os.path.isfile(nbarg):
f_in = smart_open(nbarg)
trials, best_probability, devi = f_in.readline().strip().split(
"\t")
f_in.close()
if not debug:
os.remove(nbarg)
else:
continue
for sample in self.samples:
cnvdata[sample] = CNVdata()
cnvdata[sample].trials = int(trials)
cnvdata[sample].best_probability = float(best_probability)
cnvdata[sample].min_devi = float(devi)
cnvdata[sample].ploid = int(
self.chrom_stat[sample][chrom].ploid)
cnvdata[sample].regions = list()
cnvdata[sample].data = list()
dep_data = os.path.join(
self.indir, sample, "%s.W%dS%d.fixdep.gz" %
(chrom, self.CorrectWinLen, self.CorrectShiftLen))
if not os.path.isfile(dep_data):
continue
with smart_open(dep_data) as f_in:
for line in f_in:
if line.startswith("#"):
continue
chrom, start, stop, deps = line.strip().split("\t")
start = int(start)
stop = int(stop)
deps = int(deps)
cnvdata[sample].regions.append([chrom, start, stop])
cnvdata[sample].data.append(deps)
c_stat = SaveLoad(os.path.join(self.indir, "%s.cnv.args" % chrom))
c_stat.save(cnvdata)
def run(self, debug=False):
pool = Pool(processes=cpu(use_mem=3221225472, cpu_limit=len(self.contigs)))
for chrom in self.contigs:
pool.apply_async(self.win_correct, args=(chrom,))
pool.close()
pool.join()
for chrom in self.contigs:
cnvdata = dict()
nbarg = os.path.join(self.indir, "%s.nbinom.arg" % chrom)
if os.path.isfile(nbarg):
f_in = smart_open(nbarg)
trials, best_probability, devi = f_in.readline().strip().split("\t")
f_in.close()
if not debug:
os.remove(nbarg)
else:
continue
for sample in self.samples:
cnvdata[sample] = CNVdata()
cnvdata[sample].trials = int(trials)
cnvdata[sample].best_probability = float(best_probability)
cnvdata[sample].min_devi = float(devi)
cnvdata[sample].ploid = int(self.chrom_stat[sample][chrom].ploid)
cnvdata[sample].regions = list()
cnvdata[sample].data = list()
dep_data = os.path.join(self.indir, sample,
"%s.W%dS%d.fixdep.gz" % (chrom, self.CorrectWinLen, self.CorrectShiftLen))
if not os.path.isfile(dep_data):
continue
with smart_open(dep_data) as f_in:
for line in f_in:
if line.startswith("#"):
continue
chrom, start, stop, deps = line.strip().split("\t")
start = int(start)
stop = int(stop)
deps = int(deps)
cnvdata[sample].regions.append([chrom, start, stop])
cnvdata[sample].data.append(deps)
c_stat = SaveLoad(os.path.join(self.indir, "%s.cnv.args" % chrom))
c_stat.save(cnvdata)
def bedAnalysis(**kwargs):
global pos_gc, win_gc
bed = os.path.abspath(kwargs["bed"])
reference = os.path.abspath(kwargs["reference"])
db = os.path.abspath(kwargs["db"])
outdir = os.path.abspath(kwargs["outdir"])
winlen = int(kwargs["winlen"]) if "winlen" in kwargs else 200
siftlen = int(kwargs["siftlen"]) if "siftlen" in kwargs else 20
depth_f = [os.path.abspath(i) for i in kwargs["depthfile"].split(",") if os.path.isfile(i)]
model = RegionAnalysis(reference, db)
bed_gc_out = SaveLoad(os.path.join(outdir, "win.gc"))
pos_gc_out = SaveLoad(os.path.join(outdir, "pos.gc"))
chrom_stat = SaveLoad(os.path.join(outdir, "chrom.stat"))
with smart_open(bed) as f_in:
for line in f_in:
rows = line.strip().split("\t")
chrom = str(rows[0])
if chrom not in pos_gc:
pos_gc[chrom] = dict()
start = int(rows[1])
stop = int(rows[2]) + 1
try:
model.analysis(chrom, start, stop, winlen, siftlen)
except ValueError:
continue
bed_gc_out.save(win_gc)
pos_gc_out.save(pos_gc)
chrom_stat.save(model.chrom_stat(depth_f))
model.__del__()
return bed_gc_out.fname, pos_gc_out.fname, chrom_stat.fname
def bedAnalysis(**kwargs):
global pos_gc, win_gc
bed = os.path.abspath(kwargs["bed"])
reference = os.path.abspath(kwargs["reference"])
db = os.path.abspath(kwargs["db"])
outdir = os.path.abspath(kwargs["outdir"])
winlen = int(kwargs["winlen"]) if "winlen" in kwargs else 200
siftlen = int(kwargs["siftlen"]) if "siftlen" in kwargs else 20
depth_f = [
os.path.abspath(i) for i in kwargs["depthfile"].split(",")
if os.path.isfile(i)
]
model = RegionAnalysis(reference, db)
bed_gc_out = SaveLoad(os.path.join(outdir, "win.gc"))
pos_gc_out = SaveLoad(os.path.join(outdir, "pos.gc"))
chrom_stat = SaveLoad(os.path.join(outdir, "chrom.stat"))
with smart_open(bed) as f_in:
for line in f_in:
rows = line.strip().split("\t")
chrom = str(rows[0])
if chrom not in pos_gc:
pos_gc[chrom] = dict()
start = int(rows[1])
stop = int(rows[2]) + 1
try:
model.analysis(chrom, start, stop, winlen, siftlen)
except ValueError:
continue
bed_gc_out.save(win_gc)
pos_gc_out.save(pos_gc)
chrom_stat.save(model.chrom_stat(depth_f))
model.__del__()
return bed_gc_out.fname, pos_gc_out.fname, chrom_stat.fname