def main():
usage = "usage: %prog inputfile1 inputfile2 ... inputfileN outputfile"
parser = OptionParser(usage=usage)
parser.add_option("-n", "--genename", dest="genename", help="Name of gene, prefix added to motifname.", default="")
(options, args) = parser.parse_args()
if len(args) < 3:
print ("Two args must be specified in commandline: \n" "One or more input files\n" "Output file\n")
sys.exit()
input_file_list = args[0:-1]
output_file = args[-1]
genename = options.genename
# Check output file does NOT exist
if os.path.isfile(output_file):
print "Warning: output file already exists.\n" "Press any key to overwrite file: %s" % output_file
raw_input()
print "Overwriting file: %s" % output_file
# init write file
fcount = 0
with open(output_file, "wb") as outfile:
writing_utils.write_meme_headers(outfile)
# Loop input files, reading their motifs and adding it to
for f in input_file_list:
motifname = os.path.basename(f)
# remove .txt from motifname
motifname = "".join(motifname.split(".")[:-1])
motifname = ",".join([genename, motifname, "D"])
motifs = reading_utils.read_motifs_from_file(f)
writing_utils.write_motif_to_file(outfile, motifs, motifname, end_line="\n\n")
fcount += 1
print "%s motif files crafted into: %s" % (fcount, output_file)
def main():
usage = 'usage: %prog rbp_list.txtfile rbp_db_directory '\
'meme_db_output.outputfile\n'\
'Three args must be specified in commandline: \n'\
'1) List of RBPs (ENSEMBL gene id).\n'\
'2) RBP directory containing RBP_Information_all_motifs.txt'\
' and pwms_all_motifs directory.\n'\
'3) Output file.\n'\
'-h to display this help message.\n'
parser = OptionParser(usage=usage)
parser.add_option('-f', '--dbfile', dest='rbp_filename',
help='name of RBP DB file, '\
'default is RBP_Information_all_motifs.txt',
default='RBP_Information_all_motifs.txt')
parser.add_option('-d', '--pwmdir', dest='pwm_dir',
help='Name of directory containing motif files.'\
' Default is "pwms_all_motifs".',
default='pwms_all_motifs')
(options, args) = parser.parse_args()
if len(args) < 3:
print 'Incorrect number of parameters specified.'
print usage
sys.exit()
input_rbps_path = args[0]
rbp_db_dir = args[1]
output_path = args[2]
rbp_filename = options.rbp_filename
pwm_dir = options.pwm_dir
# Check that writefile does not already exist.
# if it does, then do not overwrite and just exit
if os.path.isfile(output_path):
print '%s already exists. Aborting custom MEME db creation.' \
%output_path
sys.exit()
# Get rbp_db_path
rbp_db_path = os.path.join(rbp_db_dir, rbp_filename)
# Read list of RBPs, store as a list.
rbps_list = reading_utils.extract_column_from_textfile(input_rbps_path,
col_to_extract=0)
# Index RBP DB: {ensemblID: {RBP_Name:[], Motif_ID:[], RBP_Status:[]}}
rbp_db_dic = index_rbpdb_motifids(rbp_db_path)
# init write file
with open(output_path, 'wb') as writefile:
# Write headers
writing_utils.write_meme_headers(writefile)
# Match rbp ensemblID to indexed dic key, write motifs to file.
motifcount = 0
for rbp in rbps_list:
motifid_list, motifname_list = get_motifids_from_rbp(rbp,
rbp_db_dic)
for motifid, motifname in zip(motifid_list, motifname_list):
# Add .txt suffix to motifid
motifid_fname = ''.join([motifid, '.txt'])
motifid_path = os.path.join(rbp_db_dir, pwm_dir, motifid_fname)
# Read motifid path, get the motif.
extracted_motif = \
reading_utils.read_motifs_from_file(motifid_path,
skipheader=True,
rownames=True)
# Only write something to file if extracted_motif is not empty.
if len(extracted_motif) != 0:
# Write extracted motif to file...
writing_utils.write_motif_to_file(writefile,
extracted_motif,
motifname,
alength=4,
nsites=20)
motifcount += 1
print '%s motifs written to file: %s' %(motifcount, output_path)