def check_logic(self):
options, args = self.options, self.args
if not options.vcf_fn:
drdcommon.error("Need vcf file.")
if options.vcf_fn == '-' and not drdcommon.data_in_stdin():
drdcommon.error("No data in stdin.")
if not options.vcf_fn == '-' and not os.path.isfile(options.vcf_fn):
drdcommon.error("Vcf file does not exists.")
if options.coor_fn:
if options.coor_fn == '-' and options.vcf_fn == '-':
drdcommon.error("I cannot read two streams from stdin.")
if not os.path.isfile(options.coor_fn):
drdcommon.error("coor file does not exists.")
options.drop = True
options.coordinates_in_file = True
else:
options.coordinates_in_file = False
if len(args) == 0:
if options.wes:
self.exp_type = 'wes'
elif options.wgs:
self.exp_type = 'wgs'
else:
self.exp_type = 'null'
if not self.options.list_s_snps:
drdcommon.error("Experiment type not set.")
else:
drdcommon.error("Incorrect # of params.")
def check_logic(self):
options, args = self.options, self.args
if not options.vcf_fn:
drdcommon.error("Need vcf file.")
if options.vcf_fn == '-' and not drdcommon.data_in_stdin():
drdcommon.error("No data in stdin.")
if not options.vcf_fn == '-' and not os.path.isfile(options.vcf_fn):
drdcommon.error("Vcf file does not exists.")
if options.coor_fn:
if options.coor_fn == '-' and options.vcf_fn == '-':
drdcommon.error("I cannot read two streams from stdin.")
if not os.path.isfile(options.coor_fn):
drdcommon.error("coor file does not exists.")
options.drop = True
options.coordinates_in_file = True
else:
options.coordinates_in_file = False
if len(args) == 0:
if options.wes:
self.exp_type = 'wes'
elif options.wgs:
self.exp_type = 'wgs'
else:
self.exp_type = 'null'
if not self.options.list_s_snps:
drdcommon.error("Experiment type not set.")
else:
drdcommon.error("Incorrect # of params.")
def main():
args = parse_args()
stream = drdcommon.xopen("-")
if not drdcommon.data_in_stdin():
drdcommon.error(usage)
print Saturation(stream, args.at_least_seen).csv("\t")
stream.close()
def main():
if len(sys.argv) != 2:
drdcommon.error("Wrong # of args", usage)
if not drdcommon.data_in_stdin():
drdcommon.error("No data in stdin.", usage)
ratios_stream = drdcommon.xopen("-")
threshold = float(sys.argv[1])
CnvStateMachine(ratios_stream, threshold).run()
def main():
if len(sys.argv) != 1:
drdcommon.error("Wrong # of args", usage)
if not drdcommon.data_in_stdin():
drdcommon.error("No data in stdin.", usage)
fd_vcf = drdcommon.xopen("-")
do_work(fd_vcf)
fd_vcf.close()
def main():
if not drdcommon.data_in_stdin():
drdcommon.error("I need a data stream in stdin.", usage="-")
if not len(sys.argv) == 2:
drdcommon.error("Wrong number of parameters", usage="-")
title = sys.argv[1]
x, y = process_data(drdcommon.xopen("-"))
plot(x, y, title, xlabel="genomic window", ylabel="Average Read Depth")
def main():
if len(sys.argv) == 2:
drdcommon.error("Wrong # of args", usage)
if drdcommon.data_in_stdin() == False:
drdcommon.error("Need data in stdin.", usage)
fd_vcf = drdcommon.xopen("-")
do_work(fd_vcf)
fd_vcf.close()
def main():
if not drdcommon.data_in_stdin():
drdcommon.error("I need a data stream in stdin.", usage=_usage)
if not len(sys.argv) == 4:
drdcommon.error("Wrong number of parameters", usage=_usage)
title, _xl, _yl = sys.argv[1:]
x, y = process_data(drdcommon.xopen("-"))
plot(x, y, title, xlabel=_xl, ylabel=_yl)
def main():
if len(sys.argv) != 2:
drdcommon.error("Wrong # of args", usage)
if not drdcommon.data_in_stdin():
drdcommon.error("No data in stdin.", usage)
fd_vcf = drdcommon.xopen("-")
w_size = int(sys.argv[1])
do_work(fd_vcf, w_size)
fd_vcf.close()
def main():
if not drdcommon.data_in_stdin():
drdcommon.error("I need a data stream in stdin.", usage=_usage)
if not len(sys.argv) == 4:
drdcommon.error("Wrong number of parameters", usage=_usage)
title, _xl, _yl = sys.argv[1:]
x, y = process_data(drdcommon.xopen("-"))
plot(x, y, title, xlabel=_xl, ylabel=_yl)
def main():
if len(sys.argv) != 2:
drdcommon.error("Wrong # of args", usage)
if drdcommon.data_in_stdin() == False:
drdcommon.error("Need data in stdin.", usage)
min_num_samples = int(sys.argv[1])
fd_vcf = drdcommon.xopen("-")
do_work(fd_vcf, min_num_samples)
fd_vcf.close()
def main():
if len(sys.argv) != 2:
drdcommon.error("Wrong # of args", usage)
if not drdcommon.data_in_stdin():
drdcommon.error("No data in stdin.", usage)
windows = drdcommon.xopen("-")
bam_name = sys.argv[1]
if not os.path.isfile(bam_name):
drdcommon.error("Invalid bam file.", usage)
compute_ratios(windows, bam_name)
return prev_end + random.randint(10000, 20000)
def next_size():
return random.randint(100, 20000)
def in_n_region(start, end, ref):
num_ns, s, e = 0, int(start), int(end)
_max = e-s/8
for i in range(s, e+1):
if ref.exists(i):
num_ns += 1
if num_ns > _max:
return True
return False
if len(sys.argv) < 4 or not drdcommon.data_in_stdin():
sys.stderr.write("cat ref.fa | tool <n_events> <chrm> <chrm_size>" + "\n")
sys.exit(1)
_ = sys.argv
n_events, chrm, chrm_size = int(_[1]), _[2], int(_[3])
# Store N locations
sys.stderr.write("Loading N locations ..." + "\n")
ref = BitMask()
i = 0
for l in drdcommon.xopen("-"):
if l[0] != '-':
for c in l.rstrip():
if c.upper() == 'N':
ref.set(i)
#!/usr/bin/env python
import drdcommon
import sys
import re
def help(msg):
sys.stderr.write("ERROR: " + msg + "\n")
sys.stderr.write("Usage: cat gtf.txt | tool list_genes_names.txt > genes.coor.bed\n")
sys.exit(1)
# Main
if not drdcommon.data_in_stdin():
help("Need data in stdin")
if len(sys.argv) != 2:
help("Invalid list of arguments")
gene_names = {}
for l in drdcommon.xopen(sys.argv[1]):
name = l.split()[0]
gene_names[name] = True
drdcommon.log("%s genes loaded." % len(gene_names))
for l in drdcommon.xopen("-"):
s = l.split("\t")
if s[2] == "CDS":
chrm, start, end, _list = s[0], s[3], s[4], s[8]
g_name, e_name, t_name = None, None, None
for e in _list.split(";"):
#!/usr/bin/env python
import drdcommon
import sys
import re
def help(msg):
sys.stderr.write("ERROR: " + msg + "\n")
sys.stderr.write("Usage: cat gtf.txt | tool > exon.coor.bed\n")
sys.exit(1)
# Main
if not drdcommon.data_in_stdin():
help("Need data in stdin")
if len(sys.argv) != 1:
help("Invalid list of arguments")
for l in drdcommon.xopen("-"):
s = l.split("\t")
if s[2] == "CDS":
chrm, start, end, _list = s[0], s[3], s[4], s[8]
g_name, e_name, t_name = None, None, None
for e in _list.split(";"):
_ = e.split()
if len(_) == 2 and _[0] == "transcript_name":
t_name = re.sub('\"', '', _[1])
if len(_) == 2 and _[0] == "gene_name":
def next_size():
return random.randint(100, 20000)
def in_n_region(start, end, ref):
num_ns, s, e = 0, int(start), int(end)
_max = e - s / 8
for i in range(s, e + 1):
if ref.exists(i):
num_ns += 1
if num_ns > _max:
return True
return False
if len(sys.argv) < 4 or not drdcommon.data_in_stdin():
sys.stderr.write("cat ref.fa | tool <n_events> <chrm> <chrm_size>" + "\n")
sys.exit(1)
_ = sys.argv
n_events, chrm, chrm_size = int(_[1]), _[2], int(_[3])
# Store N locations
sys.stderr.write("Loading N locations ..." + "\n")
ref = BitMask()
i = 0
for l in drdcommon.xopen("-"):
if l[0] != '-':
for c in l.rstrip():
if c.upper() == 'N':
ref.set(i)
