def getregpot(bw, tss, output, alpha, exclude):
exclude = exclude.split(',')
left = int(exclude[0])
right = int(exclude[1])
if left == right:
_bw.getrp(bw, tss, output, alpha, 0, 0)
elif left < right:
_bw.getrp(bw, tss, output, alpha, left, right)
else:
print >> sys.stderr, "--exclude left should be smaller than right"
sys.exit(1)
rp = {}
with open(output) as inf:
for line in inf:
line = line.strip().split()
rp['\t'.join(line[:4])] = line[4]
fout = open(output, 'w')
with open(tss) as inf:
for line in inf:
line = line.strip().split()
s = line[3].split(':')
print >> fout, '\t'.join([
line[0], line[1], line[2], s[0], rp['\t'.join(line[:4])], s[1],
line[-1]
])
# use _bw to get rp for each of the files
status = 0
shape2 = 0
files = []
for key in factors:
fin = glob.glob(os.path.join(directory[int(key)], '*treat*bw'))
if fin:
print(fin[0])
# check file existence
#if not os.path.exists(os.path.basename(fin[0])+'.txt'):
if not os.path.exists(os.path.basename(fin[0]) + '_1kb.txt'):
# check status
status = os.system('bigWigInfo %s' % fin[0])
if status == 0:
##_bw.getrp(fin[0], 'histonerp/test/hg38.tss', os.path.basename(fin[0])+'.txt', 1e4, 0, 0)
_bw.getrp(fin[0], 'histonerp/test/hg38.tss',
os.path.basename(fin[0]) + '_1kb.txt', 1e3, 0, 0)
if os.path.getsize(os.path.basename(fin[0]) + '_1kb.txt') > 0:
shape2 += 1
files.append(os.path.basename(fin[0]) + '_1kb.txt')
else:
# get shape 2
if os.path.getsize(os.path.basename(fin[0]) + '_1kb.txt') > 0:
shape2 += 1
files.append(os.path.basename(fin[0]) + '_1kb.txt')
f = h5py.File('margeRP_%s_1kb.h5' % factor, 'a')
if not ('RP' in f.keys()):
RP = f.create_dataset("RP",
dtype=np.float32,
shape=(l, shape2),
import h5py, _bw
import numpy as np
import os
from pkg_resources import resource_filename
import gzip
meta = resource_filename("_bw", "%s.tss" % (snakemake.params.sp))
_bw.getrp(snakemake.input.bw, meta, snakemake.output.rp, 1e4, 0, 0)
# single sample RP to HDF5
tmp = []
n = 0
with open(snakemake.output.rp) as inf:
for line in inf:
n += 1
line = line.split()[4]
tmp.append(float(line.strip()))
tmp = np.array(tmp, dtype='float32')
with h5py.File(snakemake.output.h5, 'a') as store:
RP = store.create_dataset("RP",
dtype=np.float32,
shape=(n, 1),
compression='gzip',
shuffle=True,
fletcher32=True)
RP[:, 0] = tmp
store.flush()
os.system('bigWigAverageOverBed %s %s stdout > %s' %
(snakemake.input.bw, snakemake.params.bin1kb, snakemake.output.ct))
# check status
status = os.system('bigWigInfo %s' % fin[0])
if status == 0:
# without promoter +/- 1000bp
#if factors[key] == 'H3K4me3':
# _bw.getrp(fin[0], 'histonerp/test/mm10.tss', os.path.basename(fin[0])+'_rdown1kb.txt', 1e3, -1000, 1000)
#else:
# _bw.getrp(fin[0], 'histonerp/test/mm10.tss', os.path.basename(fin[0])+'_rdown1kb.txt', 1e4, -1000, 1000)
# with promoter +/- 1000bp
#if factors[key] == 'H3K4me3':
# _bw.getrp(fin[0], 'histonerp/test/mm10.tss', os.path.basename(fin[0])+'_rdown1kb.txt', 1e3, 0, 0)
#else:
# _bw.getrp(fin[0], 'histonerp/test/mm10.tss', os.path.basename(fin[0])+'_rdown1kb.txt', 1e4, 0, 0)
_bw.getrp(fin[0], '../MARGEData/histonerp/test/mm10.tss',
os.path.basename(fin[0]) + '.txt', 1e4, 0, 0)
if os.path.getsize(os.path.basename(fin[0]) + '.txt') > 0:
shape2 += 1
files.append(os.path.basename(fin[0]) + '.txt')
else:
# get shape 2
if os.path.getsize(os.path.basename(fin[0]) + '.txt') > 0:
shape2 += 1
files.append(os.path.basename(fin[0]) + '.txt')
f = h5py.File('margeRP_%s_mm.h5' % factor, 'a')
if not ('RP' in f.keys()):
RP = f.create_dataset("RP",
dtype=np.float32,
shape=(l, shape2),