#!/usr/bin/env python
import csv
from collections import defaultdict
import sys, getopt
import os
import math
from shutil import copyfile
import pandas as pd
from scipy.stats import chi2
from Bio import SeqIO
import os.path
SOFTWARES = {'meme','dreme','centrimo','meme_tomtom'}
##TYPES_SEARCHES = {'All','Narrow','Const','Max'}
def get_size_fasta(path_motif, exp_design_name):
with open(path_motif + 'Fasta_Summary.txt', 'w') as fasta_summary:
for data_name in os.listdir(path_motif+'fasta/'):
if data_name.endswith('.fasta'):
print(data_name)
fasta_seq = SeqIO.to_dict(SeqIO.parse(path_motif+'fasta/'+data_name, "fasta"))
print(data_name,len(fasta_seq))
fasta_summary.write(data_name.replace('.fasta','')+'\t'+str(len(fasta_seq))+'\n')
def parse_all_logs(path_motif, exp_design_name):
with open(path_motif + '../Motif_search.log', 'w') as general_log:
header = 'Data_name\t' + '\t'.join(SOFTWARES)
general_log.write(header+'\n')
for data_name in os.listdir(path_motif):
if not data_name.startswith('.') and not data_name.endswith('.log') and not data_name.endswith('.sh'):
progress_filename = path_motif + data_name + '/progress_log.txt'
line = data_name + '\t'
for software in SOFTWARES:
found = False
with open(progress_filename,'r') as file:
for row in file:
status_soft = 'name: '+software+' status: 0'
if status_soft in row:
found = True
if found:
line += '1\t'
else:
line += '0\t'
general_log.write(line.strip() + '\n')
def parse_dreme(path_motif, exp_design_name):
with open(path_motif + 'dreme.sh', 'w') as dreme_file:
for data_name in os.listdir(path_motif):
if exp_design_name in data_name:
print(data_name)
if not data_name.startswith('.') and not data_name.endswith('.log') and not data_name.endswith('.sh'):
# print(data_name)
dreme_xml = path_motif + data_name + '/dreme_out/dreme.xml'
if os.path.exists(dreme_xml):
dreme_list = list()
with open(dreme_xml, 'r') as file:
for row in file:
dreme_list.append(row.strip())
if '<command_line>' in row:
dreme_file.write(row)
dreme_txt = path_motif + data_name + '/dreme_out/dreme.txt'
dreme_txt_list = list()
with open(dreme_txt, 'r') as file:
for row in file:
dreme_txt_list.append(row.strip())
dreme_txt_intro = ''
write_intro = False
for row in dreme_txt_list:
if row.startswith('MOTIF '):
break
if 'MEME version' in row:
write_intro = True
if write_intro:
dreme_txt_intro += row + '\n'
#print(dreme_txt_intro)
# parse dreme
with open(path_motif + '../logs/Dreme_' + data_name + '.log', 'w') as general_log:
header = 'ID\tMotif\tNb_sites\tPos_Occurence\tNeg_Occurence\tNb_Seq\t' \
'Percent\tPos_Percent\tNeg_Percent\tEvalue'
general_log.write(header + '\n')
# search length fasta
for line in dreme_list:
if '<positives name=' in line:
number_seq = float(line.split(' ')[2].replace('\"','').replace('count=',''))
# search motifs
for i in range(1, len(dreme_list)):
line = dreme_list[i]
if '<motifs>' in line:
index_motifs = i
break
# get motifs info
for k in range(index_motifs+1, len(dreme_list)):
if '<motif' in dreme_list[k]:
new_line = dreme_list[k].replace('\"','').split(' ')
#print(new_line)
id = new_line[1].replace('id=', '')
#print(id,data_name)
print(dreme_txt_intro)
seq = new_line[2].replace('seq=', '')
occ = float(new_line[4].replace('nsites=', ''))
pos_occ = float(new_line[5].replace('p=', ''))
neg_occ = float(new_line[6].replace('n=', ''))
evalue = new_line[8].replace('evalue=', '')
# save table
line_to_write = [id, seq, str(occ), str(pos_occ),str(neg_occ), str(number_seq),
str(occ/number_seq), str(pos_occ/number_seq), str(neg_occ/number_seq),
str(evalue)]
general_log.write('\t'.join(line_to_write) + '\n')
# copy png file
for imagefile in os.listdir(path_motif + data_name + '/dreme_out/'):
if id in imagefile:
if 'nc_' in imagefile:
src = path_motif + data_name + '/dreme_out/' + imagefile
dst = path_motif + '../motif_figure/'+seq + '_nc.png'
copyfile(src, dst)
else:
src = path_motif + data_name + '/dreme_out/' + imagefile
dst = path_motif + '../motif_figure/' + seq + '_rc.png'
copyfile(src, dst)
# save motif
with open(path_motif + '../motif/' + seq + '.meme', 'w') as meme_file:
meme_file.write(dreme_txt_intro+'\n')
# find motif info
dreme_txt_motif = ''
write_motif = False
for row in dreme_txt_list:
if write_motif and row.startswith('MOTIF '):
break
if ('MOTIF '+seq+' DREME') in row:
write_motif = True
if write_motif:
dreme_txt_motif += row + '\n'
print(dreme_txt_motif)
meme_file.write(dreme_txt_motif + '\n')
def regroup_motifs(path_motif, exp_design_name):
motif_set = set()
dict_dataset = defaultdict(list)
dict_evalue = defaultdict(list)
for data_name in os.listdir(path_motif+'logs/'):
if data_name.startswith('Dreme_') and data_name.endswith('.log'):
type_data = data_name.replace('Dreme_','').replace('.log','')
print(type_data,data_name)
with open(path_motif+'logs/'+data_name,'r') as log_file:
log_file.readline()
for line in log_file:
motif = line.split('\t')[1]
evalue = line.strip().split('\t')[-1]
motif_set.add(motif)
dict_dataset[motif].append(type_data)
dict_evalue[motif].append(evalue)
with open(path_motif + 'Motif_'+exp_design_name+'.txt','w') as motif_log:
motif_log.write("Motif\tData\tP-value\n")
for motif in motif_set:
motif_log.write(motif+'\t'+';'.join(dict_dataset[motif])+'\t'+';'.join(dict_evalue[motif])+'\n')
def regroup_figures(path_motif, motif_list):
'''
Create SVG file with all motifs included
:param path_motif:
:param motif_list:
:return:
'''
print('Create PNG file with all motifs')
df_summary = pd.read_csv(path_motif + motif_list + '.txt', sep='\t')
df_summary = df_summary.sort_values(by=['Motif'], ascending=[True])
svg_text = '<?xml version=\"1.0\" encoding=\"utf-8\"?>\n<!-- Generator: Adobe Illustrator 15.0.0, SVG Export ' \
'Plug-In . SVG Version: 6.00 Build 0) -->\n<!DOCTYPE svg PUBLIC \"-//W3C//DTD SVG 1.1//EN\" \"' \
'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd\">\n<svg version=\"1.1\" id=\"Calque_1\" ' \
'xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\" x=\"0px\" y=\"0px\"' \
' width=\"1300px\" height=\"'+str(len(df_summary.index)*100)+'px\" viewBox=\"0 0 1300 '\
+str(len(df_summary.index)*100)+'\" xml:space=\"preserve\">\n'
i=0
for index, row in df_summary.iterrows():
motif = row['Motif']
data = row['Data']
pvalue = row['P-value']
image_file = 'motif_figure/' + motif + '_nc.png'
new_row = '<image overflow=\"visible\" width=\"299\" height=\"176\" xlink:href=\"'+image_file+'\" ' \
'transform=\"matrix(0.5184 0 0 0.5184 10.00 '+ str(float(100*i))+')\">\n</image>\n'
image_file = 'motif_figure/' + motif + '_rc.png'
new_row += '<image overflow=\"visible\" width=\"299\" height=\"176\" xlink:href=\"' + image_file + '\" ' \
'transform=\"matrix(0.5184 0 0 0.5184 170.00 '+ str(float(100*i))+')\">\n</image>\n'
new_row += '<text transform=\"matrix(0.5184 0 0 0.5184 340 '+ str(float(50+100*i))+')\" font-family=' \
'\"\'MyriadPro-Regular\'\" font-size=\"40\">'+data+'</text>\n'
new_row += '<text transform=\"matrix(0.5184 0 0 0.5184 800 '+ str(float(50+100*i))+')\" font-family=' \
'\"\'MyriadPro-Regular\'\" font-size=\"40\">'+pvalue+'</text>\n'
svg_text += new_row
i+=1
svg_text += '</svg>\n'
with open(path_motif + motif_list + '.svg', 'w') as figure_file:
figure_file.write(svg_text)
print('Convert svg to png')
os.system('convert '+path_motif + motif_list + '.svg '+path_motif + motif_list + '.png')
os.system('convert '+path_motif + motif_list + '.svg '+path_motif + '../../All_Figures/Motif/' + motif_list + '.png')
def run_fimo(path_motif, path_script, exp_design_name, motif_list_filename, data_list_filename):
motif_list = list()
with open(path_motif + motif_list_filename + '.txt', 'r') as motif_file:
for row in motif_file:
motif_list.append(row.split('\t')[0].strip())
data_list = list()
with open(path_motif + data_list_filename + '.txt', 'r') as data_file:
for row in data_file:
data_list.append(row.split('\t')[0].strip())
with open(path_motif + 'Fimo.sh', 'w') as motif_sh:
for motif in motif_list:
motif_file = path_script + 'motif/' + motif + '.meme'
print(motif_file)
for data_name in data_list:
fasta_file = path_script + 'fasta/' + data_name + '.fasta'
folder_fimo = path_script + 'fimo/' + motif + '_' + data_name
fimo_sh = 'fimo --parse-genomic-coord --verbosity 1 --thresh 1e-2 --max-stored-scores 1000000 '\
'--oc ' + folder_fimo + ' ' + motif_file + ' ' + fasta_file
print(fimo_sh)
motif_sh.write(fimo_sh+'\n')
def run_centrimo(path_motif, path_script, exp_design_name, motif_list_filename, data_list_filename):
motif_list = list()
with open(path_motif + motif_list_filename + '.txt', 'r') as motif_file:
for row in motif_file:
motif_list.append(row.split('\t')[0].strip())
data_list = list()
with open(path_motif + data_list_filename + '.txt', 'r') as data_file:
for row in data_file:
data_list.append(row.split('\t')[0].strip())
with open(path_motif + 'Centrimo.sh', 'w') as motif_sh:
for motif in motif_list:
motif_file = path_motif + 'motif/' + motif + '.meme'
print(motif_file)
for data_name in data_list:
fasta_file = path_motif + 'fasta/' + data_name + '.fasta'
folder_centrimo = path_motif + 'centrimo/' + motif + '_' + data_name
centrimo_sh = 'centrimo -seqlen 100 -verbosity 1 -oc ' + folder_centrimo + ' -score 5.0 '\
'-ethresh 10.0 ' + fasta_file + ' '+motif_file
print(centrimo_sh)
motif_sh.write(centrimo_sh + '\n')
def parse_fimo(path_motif, exp_design_name, motif_list_filename, data_list_filename):
motif_list = list()
with open(path_motif + motif_list_filename + '.txt', 'r') as motif_file:
for row in motif_file:
motif_list.append(row.split('\t')[0].strip())
data_list = list()
with open(path_motif + data_list_filename + '.txt', 'r') as data_file:
for row in data_file:
data_list.append(row.split('\t')[0].strip())
fasta_to_length = dict()
with open(path_motif + 'Fasta_Summary.txt', 'r') as fasta_summary:
for row in fasta_summary:
fasta_to_length[row.split('\t')[0]] = int(row.strip().split('\t')[1])
with open(path_motif + 'Fimo_Summary_Max.txt', 'w') as fimo_summary_max, \
open(path_motif + 'Fimo_Summary_All.txt', 'w') as fimo_summary_all:
headers = 'Data_Name\tSeq\tData\tTotal_Occurence\tNb_Peaks\tNb_Peaks_Fasta\tRatio\n'
fimo_summary_max.write(headers)
fimo_summary_all.write(headers)
for seq in motif_list:
for data_type in data_list:
data_name = seq + '_'+ data_type
print(data_name)
if not os.path.exists(path_motif + 'fimo/' + data_name + '/fimo.txt'):
print('Cannot find:' + path_motif + 'fimo/' + data_name + '/fimo.txt')
print('Create void table to proceed motif algorithm')
with open(path_motif + 'fimo_template.txt', 'r') as fimo_file, \
open(path_motif + 'fimo/' + data_name + '/fimo.txt', 'w') as fimo_table:
fimo_table.write(fimo_file.readline())
fimo_table.write(fimo_file.readline())
with open(path_motif + 'fimo/' + data_name + '/fimo.txt','r') as fimo_file, \
open(path_motif + 'fimo/' + data_name + '/fimo_table.txt', 'w') as fimo_table:
fimo_table.write('Peak\tOcc\tchi_fisher\tscore\n')
fimo_file_csv = csv.DictReader(fimo_file, delimiter='\t')
peak_to_value = defaultdict(list)
peak_to_occurence = defaultdict(int)
for row in fimo_file_csv:
value = float(row['score'])
if value > 0:
peak_to_occurence[row['sequence name']] += 1
peak_to_value[row['sequence name']].append(str(row['score']))
total_occ = 0
for peak, occ in peak_to_occurence.items():
chi_fisher = 0
for pvalue in peak_to_value[peak]:
#chi_fisher += -2 * math.log1p(float(pvalue))
chi_fisher += float(pvalue)
#print chi_fisher
fimo_table.write(peak+'\t'+str(occ)+'\t'+str(chi_fisher)+'\t'+';'.join(peak_to_value[peak])+'\n')
total_occ += occ
seq = data_name[0:data_name.index('_')]
fasta_name = data_name[data_name.index('_')+1:len(data_name)]
fasta_length = fasta_to_length[fasta_name]
new_line = data_name + '\t' + seq + '\t' + fasta_name + '\t' + str(total_occ) + '\t' \
+ str(len(peak_to_occurence)) + '\t' + str(fasta_length) + '\t' + \
str(total_occ) + '\n'
if 'All_' in data_name:
fimo_summary_all.write(new_line)
else:
fimo_summary_max.write(new_line)
def create_motif_table(path_motif, exp_design_name, path_peaks, motif_list_filename, data_list_filename, type_search):
set_seq = set()
set_type_data = set()
df_summary = pd.read_csv(path_motif + 'Fimo_Summary_' + type_search + '.txt', index_col=0, sep='\t')
motif_list = list()
with open(path_motif + motif_list_filename + '.txt', 'r') as motif_file:
for row in motif_file:
motif_list.append(row.split('\t')[0].strip())
data_list = list()
with open(path_motif + data_list_filename + '.txt', 'r') as data_file:
for row in data_file:
data_type = row.split('\t')[0].strip()
data_list.append(data_type)
set_peaks = set()
df_peaks = pd.read_csv(path_peaks+ exp_design_name + '_Peaks.txt', index_col=0, sep='\t')
for index in df_peaks.index:
set_peaks.add(index)
for seq in motif_list:
print(seq)
with open(path_motif + 'mutual_information/Fimo_Table_' + type_search + '_' + seq + '.txt', 'w') as fimo_table, \
open(path_motif + 'mutual_information/Fimo_Table_' + type_search + '_' + seq + '_Count.txt', 'w') as fimo_summary:
header_type_data = list(data_list)
#header_type_data[0] = 'All'
fimo_summary.write('Pres_Abs\t' + '\t'.join(header_type_data)+'\n')
fimo_table.write('Peak\t' + '\t'.join(header_type_data) + '\n')
peak_to_occ = defaultdict(dict)
data_to_count = dict()
for type_data in data_list:
print(type_data)
data_name = seq + '_' + type_data
df_motif = pd.read_csv(path_motif + 'fimo/' + data_name + '/fimo_table.txt', index_col=0, sep='\t')
count_occ = 0
for peak in set_peaks:
if peak in df_motif.index:
peak_to_occ[peak][type_data] = df_motif['chi_fisher'][peak]
count_occ += df_motif['chi_fisher'][peak]
else:
peak_to_occ[peak][type_data] = 0
data_to_count[type_data] = count_occ
new_line = 'Present\t'
for index in data_list:
new_line += str(data_to_count[index]) + '\t'
fimo_summary.write(new_line.strip() + '\n')
for peak in peak_to_occ:
new_line = peak + '\t'
for index in data_list:
new_line += str(peak_to_occ[peak][index]) + '\t'
fimo_table.write(new_line.strip() + '\n')
def create_occurence_table(path_motif, exp_design_name, motif_list_filename, data_list_filename, type_search):
fasta_to_length = dict()
with open(path_motif + 'Fasta_Summary.txt', 'r') as fasta_summary:
for row in fasta_summary:
fasta_to_length[row.split('\t')[0]] = int(row.strip().split('\t')[1])
motif_list = list()
with open(path_motif + motif_list_filename + '.txt', 'r') as motif_file:
for row in motif_file:
motif_list.append(row.split('\t')[0].strip())
data_list = list()
with open(path_motif + data_list_filename + '.txt', 'r') as data_file:
for row in data_file:
data_type = row.split('\t')[0].strip()
data_list.append(data_type)
with open(path_motif + 'Fimo_Table_' + type_search + '.txt', 'w') as fimo_summary:
header = 'Sequence\t' + '\t'.join(list(data_list))
fimo_summary.write(header + '\n')
for seq in motif_list:
print(seq)
df_motif = pd.read_csv(path_motif + 'mutual_information/Fimo_Table_' + type_search + '_' + seq + '_Count.txt'
, index_col=0, sep='\t')
new_line = seq + '\t'
print(df_motif.columns)
for type_data in data_list:
row_name = seq + '_' + type_search + '_' + exp_design_name + type_data
length = fasta_to_length[type_data]
print(length)
if type_data == '':
type_data = 'All'
value = df_motif[type_data]['Present']
if length != 0 :
new_line += str(float(value)/float(length)) + '\t'
else:
new_line += str(0) + '\t'
fimo_summary.write(new_line.strip() + '\n')
#exp_design_name = 'LivOld'
exp_design_name = 'CecAm_MaxMaxValues'
#path = '/Users/cbecavin/Documents/m6aAkker/'
path = '/Volumes/m6aAkker/'
path_motif = path + 'PeakDiffExpression/' + exp_design_name + '/Motif/'
path_peaks = path + 'PeakDetection/Peaks/'
if not os.path.isdir(path_motif + 'logs/'):
os.mkdir(path_motif + 'logs/')
if not os.path.isdir(path_motif + 'motif/'):
os.mkdir(path_motif + 'motif/')
if not os.path.isdir(path_motif + 'motif_figure/'):
os.mkdir(path_motif + 'motif_figure/')
if not os.path.isdir(path_motif + 'fimo/'):
os.mkdir(path_motif + 'fimo/')
#path_script = path + 'PeakDiffExpression/' + exp_design_name + '/Motif/'
path_script = '/pasteur/projets/policy01/m6aAkker/' + 'PeakDiffExpression/' + exp_design_name + '/Motif/'
# get size fasta
#get_size_fasta(path_motif, exp_design_name)
# first look at log to see which software crashed
#parse_all_logs(path_motif+'/results/', exp_design_name)
# Parse meme results and extract all motifs
#parse_dreme(path_motif+'/results/', exp_design_name)
# regroup all motifs
#regroup_motifs(path_motif, exp_design_name)
motif_list_filename = 'Motif_'+exp_design_name+'_Filter'
# put all motifs figures together to filter the list
#regroup_figures(path_motif, motif_list_filename)
data_list_filename = 'list_data_' + exp_design_name
#data_list_filename = 'list_data_All'
# write sh file for running fimo
run_fimo(path_motif, path_script, exp_design_name, motif_list_filename, data_list_filename)
# write sh file for running centrimo
#run_centrimo(path_motif, path_script, exp_design_name, motif_list_filename, data_list_filename)
# parse fimo results
#parse_fimo(path_motif, exp_design_name, motif_list_filename, data_list_filename)
#motif_list_filename = 'list_motifs_All'
#motif_list_filename = 'list_motifs_Max'
#data_list_filename = 'list_data_All'
#data_list_filename = 'list_data_Max'
#type_search = 'All'
#type_search = 'Max'
# create a table for each motif of occurence in list
#create_motif_table(path_motif, exp_design_name, path_peaks, motif_list_filename, data_list_filename, type_search)
# create a table for occurence of each motif
#create_occurence_table(path_motif, exp_design_name, motif_list_filename, data_list_filename, type_search)