Homec4science
Diffusion scATAC-seq (master) Browse

scATAC-seq/scripts/pwm_tools/extract_transfac_pwm.py
92b4e33930a3master

Last Change
Raw File

extract_transfac_pwm.py
View Options

import optparse
import os
import re
def parse_file(file_in, norm, pseudocount):
# matrix in vertical format
matrix_v = list()
with open(file_in) as f_in:
# if currently reading the matrix
matrix = False
for line in f_in:
line = line.rstrip()
# before matrix
if matrix is False:
if line.startswith("P0") or line.startswith("PO"):
matrix = True
# inside matrix
else:
# end of matrix
if re.match(r"^\d+", line) is None:
# if line.startswith("XX"):
matrix = False
break
# inside matrix
else:
values = re.split(r"\s+", line)[1:]
values = [float(x)+pseudocount for x in values]
if norm:
tot = sum(values)
values = [x/tot for x in values]
matrix_v.append(values)
# matrix in vertical format
return matrix_v
if __name__ == "__main__":
# parse options
usage = "usage: %s [options]" % os.path.basename(__file__)
epilog = "This program reads a transfac PWM file returns the PWM." \
"Written by Romain Groux, August 2019"
parser = optparse.OptionParser(usage=usage, epilog=epilog)
parser.add_option("-i", "--input", dest="file_in", default=None, type="string", action="store",
help="The addresse of the transfac PWM file.")
parser.add_option("--norm", dest="norm", action="store_true",
help="Whether the values should be normalized to probabilities.")
(options, args) = parser.parse_args()
file_in = options.file_in
norm = options.norm
# matrix in vertical format
matrix_v = parse_file(file_in, norm, 1)
# matrix in horizontal format
nrow = 4
ncol = len(matrix_v)
matrix_h = ["" for _ in range(0, nrow, 1)]
for i in range(0, nrow, 1):
for j in range(0, ncol, 1):
matrix_h[i] += "%.4f\t" % matrix_v[j][i]
matrix_h[i] = matrix_h[i][:-1]
# print matrix in horizontal format
print('\n'.join(matrix_h))
c4science · Help