def main():
if len(sys.argv) != 4:
drdcommon.error("Wrong # of args", usage)
pattern = sys.argv[1]
re_id = sys.argv[2] # Regular expression to extract id
f_dir = sys.argv[3]
l_ids = []
l = lambda:defaultdict(l)
h = l() # hold all data in mem
files_to_iterate = drdcommon.files_in_dir(f_dir, pattern)
sys.stderr.write("# of files to process: " + str(len(files_to_iterate)) + "\n")
for fn in files_to_iterate:
try:
sid = re.search(re_id, fn).group(1)
l_ids.append(sid)
except:
raise(Exception('Problems extracting id using regular expression.'))
load_data(sid, fn, h)
# print header
out("chrm start ")
for _id in l_ids:
out("%s " % _id)
out("\n")
for chrm, one in h.items():
for start, two in one.items():
out("%s %s " % (str(chrm), str(start)))
for sid, nr in two.items():
out(str(nr[1]) + " ")
out("\n")
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 __init__(self, h):
self.set_attr(h)
if self.id == '':
common.error('I cannot create a bam without id ')
if not self.valid_path():
common.error('Invalid path [%s] for bam with id = [%s]' %
(self.path, self.id))
def main():
if len(sys.argv) == 1:
fd_reads = drdcommon.xopen("-")
do_work(fd_reads)
fd_reads.close()
else:
drdcommon.error("Incorrect # of params.", usage)
def main():
if len(sys.argv) != 3:
drdcommon.error("Wrong # of args", usage)
pattern = sys.argv[1]
re_id = sys.argv[2] # Regular expression to extract id
l = lambda: defaultdict(l)
h = l()
l_ids = []
for fn in drdcommon.files_in_dir(".", pattern):
try:
sid = re.search(re_id, fn).group(1)
l_ids.append(sid)
except:
raise (
Exception('Problems extracting id using regular expression.'))
load_data(sid, fn, h)
# print header
sys.stdout.write("chrm coor type num_over_zero ")
for _id in l_ids:
sys.stdout.write("%s " % _id)
print ""
for t, one in h.items():
for chrm, two in one.items():
for coor, three in two.items():
sys.stdout.write("%s %s %s %s " %
(str(chrm), str(coor), t, len(three)))
for sid in l_ids:
if sid in three:
sys.stdout.write(str(three[sid]) + " ")
else:
sys.stdout.write("0 ")
print ""
def main():
if len(sys.argv) == 1:
fd_reads = drdcommon.xopen("-")
do_work(fd_reads)
fd_reads.close()
else:
drdcommon.error("Incorrect # of params.", usage)
def main():
if len(sys.argv) == 5:
fn_sam, fn_hits = sys.argv[1:3]
min_mapq, pm_hits = [int(i) for i in sys.argv[3:]]
ValidateChip(fn_sam, fn_hits, min_mapq, pm_hits).do_work()
else:
drdcommon.error("Incorrect # of params.", usage)
def main():
if len(sys.argv) == 5:
fn_sam, fn_hits = sys.argv[1:3]
min_mapq, pm_hits = [ int(i) for i in sys.argv[3:]]
ValidateChip(fn_sam, fn_hits, min_mapq, pm_hits).do_work()
else:
drdcommon.error("Incorrect # of params.", usage)
def main():
if len(sys.argv) != 3:
drdcommon.error("Wrong # of args", usage)
pattern = sys.argv[1]
re_id = sys.argv[2] # Regular expression to extract id
l = lambda:defaultdict(l)
h = l()
l_ids = []
for fn in drdcommon.files_in_dir(".", pattern):
try:
sid = re.search(re_id, fn).group(1)
l_ids.append(sid)
except:
raise(Exception('Problems extracting id using regular expression.'))
load_data(sid, fn, h)
# print header
sys.stdout.write("chrm coor type num_over_zero ")
for _id in l_ids:
sys.stdout.write("%s " % _id)
print ""
for t, one in h.items():
for chrm, two in one.items():
for coor, three in two.items():
sys.stdout.write("%s %s %s %s " % (str(chrm), str(coor), t, len(three)))
for sid in l_ids:
if sid in three:
sys.stdout.write(str(three[sid]) + " ")
else:
sys.stdout.write("0 ")
print ""
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 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 check_input():
if len(sys.argv) != 3:
drdcommon.error("Wrong # of args", usage)
bam_fn, probes_fn = sys.argv[1], sys.argv[2]
logging.info("bam: %s probes: %s" % (bam_fn, probes_fn))
if not os.path.isfile(bam_fn):
drdcommon.error("Invalid bam file.", usage)
return bam_fn, probes_fn
def __load_vcf(self):
self.vcf = Vcf(self.fd_vcf)
self.vcf.load_meta_header()
if self.drop and (not self.coordinates_in_file
and self.vcf.num_of_samples < 2):
drdcommon.error(
"I need a population level vcf in order to drop species snps.")
def main():
if len(sys.argv) == 2:
pattern = sys.argv[1]
data = load_data(pattern)
boxplot(data, title=pattern, y_limit=50)
else:
drdcommon.error("Incorrect # of params.", usage)
def main():
if len(sys.argv) == 3:
df = pd.read_table(sys.argv[1])
title = "MAF CRV"
labels = ["0-0.05", "0.05-0.1", "0.1-0.15", "0.15-0.2", "0.2-0.25", "0.25-0.3", "0.3-0.35", "0.35-0.4", "0.4-0.45", "0.45-0.5" ]
drdplots.barplot(df.counts, labels, title, ofn=sys.argv[2])
else:
drdcommon.error("Wrong number of args. Need input tsv file and output png.")
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 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 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) == 4:
fd_vcf = drdcommon.xopen("-")
fd_csv = drdcommon.xopen(sys.argv[1])
do_work(fd_vcf, fd_csv)
fd_vcf.close()
fd_csv.close()
else:
drdcommon.error("Incorrect # of params.", usage)
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 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 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) == 3:
logratios = process_data(drdcommon.xopen("-"))
bin_nums = range(1, len(logratios)+1)
title = sys.argv[1]
output_fn = sys.argv[2]
plot(output_fn,
bin_nums, logratios, title, xlabel="bin #", ylabel="log2ratios (sample/control)")
else:
drdcommon.error("Wrong number of args. <title> <output.filename>")
def main():
if len(sys.argv) == 3:
fd_vcf = drdcommon.xopen("-")
fd_pheno_tsv = drdcommon.xopen(sys.argv[1])
fd_haplo_tsv = drdcommon.xopen(sys.argv[2])
do_work(fd_vcf, fd_pheno_tsv, fd_haplo_tsv)
fd_vcf.close()
fd_pheno_tsv.close()
fd_haplo_tsv.close()
else:
drdcommon.error("Incorrect # of params.", usage)
def main():
if len(sys.argv) == 1:
fd = drdcommon.xopen("-")
std, counts = process_data(fd)
title = "std dev freq of var allele ratios"
drdplots.scatter_plot("std.dist.png",
std, log_it(counts, 10),
title=title, xlabel="std deviation",
ylabel="log10(counts)", dot_size=10)
fd.close()
else:
drdcommon.error("Wrong number of args. Just need std values in stdin.")
def main():
if len(sys.argv) == 3:
logratios = process_data(drdcommon.xopen("-"))
bin_nums = range(1, len(logratios) + 1)
title = sys.argv[1]
output_fn = sys.argv[2]
plot(output_fn,
bin_nums,
logratios,
title,
xlabel="bin #",
ylabel="log2ratios (sample/control)")
else:
drdcommon.error("Wrong number of args. <title> <output.filename>")
def main():
if len(sys.argv) == 1:
fd = drdcommon.xopen("-")
std, counts = process_data(fd)
title = "std dev freq of var allele ratios"
drdplots.scatter_plot("std.dist.png",
std,
log_it(counts, 10),
title=title,
xlabel="std deviation",
ylabel="log10(counts)",
dot_size=10)
fd.close()
else:
drdcommon.error("Wrong number of args. Just need std values in stdin.")
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)
def main():
if len(sys.argv) != 4:
drdcommon.error("Wrong # of args", usage)
n_reads, bam_name, target_chrm = int(sys.argv[1]), sys.argv[2], sys.argv[3]
if not os.path.isfile(bam_name):
drdcommon.error("Invalid bam file.", usage)
samfile = pysam.Samfile(bam_name, "rb")
chrms, chrm_name_to_length = gen_chrm_lenghts(samfile)
if not target_chrm in chrm_name_to_length:
samfile.close()
drdcommon.error("Chrm not present in bam header.", usage)
print_bins(samfile, n_reads, bam_name, target_chrm, chrm_name_to_length)
samfile.close()
def main():
if len(sys.argv) != 4:
drdcommon.error("Wrong # of args", usage)
n_reads, bam_name, target_chrm = int(sys.argv[1]), sys.argv[2], sys.argv[3]
if not os.path.isfile(bam_name):
drdcommon.error("Invalid bam file.", usage)
samfile = pysam.Samfile(bam_name, "rb")
chrms, chrm_name_to_length = gen_chrm_lenghts(samfile)
if not target_chrm in chrm_name_to_length:
samfile.close()
drdcommon.error("Chrm not present in bam header.", usage)
print_bins(samfile, n_reads, bam_name, target_chrm, chrm_name_to_length)
samfile.close()
def main():
if len(sys.argv) == 3:
sam_fn, pattern_fn_hits = sys.argv[1:]
JoinData(sam_fn, pattern_fn_hits).do_work()
else:
drdcommon.error("Incorrect # of params.", usage)
def __init__(self, h):
self.set_attr(h)
if self.id == '':
common.error('I cannot create a bam without id ')
if not self.valid_path():
common.error('Invalid path [%s] for bam with id = [%s]' % (self.path, self.id))
def __init__(self, fn):
self.df = pd.read_table(fn)
for c in self.MANDATORY_COLS:
if c not in self.df.columns:
common.error("I couldn't find column [%s] in project tsv" % c)
def main():
if len(sys.argv) == 2:
pattern = sys.argv[1]
do_work(pattern)
else:
drdcommon.error("Incorrect # of params.", usage)
def main():
if len(sys.argv) == 2:
pattern = sys.argv[1]
do_work(pattern)
else:
drdcommon.error("Incorrect # of params.", usage)
def __load_vcf(self):
self.vcf = Vcf(self.fd_vcf)
self.vcf.load_meta_header()
if self.drop and (not self.coordinates_in_file and self.vcf.num_of_samples < 2):
drdcommon.error("I need a population level vcf in order to drop species snps.")
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 __init__(self, fn):
self.df = pd.read_table(fn)
for c in self.MANDATORY_COLS:
if c not in self.df.columns:
common.error("I couldn't find column [%s] in project tsv" % c)
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():
if len(sys.argv) == 3:
sam_fn, pattern_fn_hits = sys.argv[1:]
JoinData(sam_fn, pattern_fn_hits).do_work()
else:
drdcommon.error("Incorrect # of params.", usage)
