Skip to content
Snippets Groups Projects
parse_motif_search.py 20.3 KiB
Newer Older
#!/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
import numpy as np
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 add_motif(path_motif, motif, motif_list_filename):
    motif_file = path_motif + '/motif/' + motif + '.meme'
Christophe  BECAVIN's avatar
Christophe BECAVIN committed
    motif_figure_file = path_motif + '/motif_figure/' + motif + '_nc.eps'
#    motif_figure_file = path_motif + '/motif_figure/' + motif + '_nc.png'
    # Meme-suite as to be installed for using = iupac2meme
    iupac2_command = 'iupac2meme -dna ' + motif + ' > ' + motif_file
    print(iupac2_command)
Christophe  BECAVIN's avatar
Christophe BECAVIN committed
    os.system(iupac2_command)
    ceqlogo_command = 'ceqlogo -i ' + motif_file + ' -m ' + motif + ' -o ' + motif_figure_file + ' -f eps'
#    ceqlogo_command = 'ceqlogo -i ' + motif_file + ' -m ' + motif + ' -o '+motif_figure_file+' -f png'
    print('Run ' + ceqlogo_command)
    os.system(ceqlogo_command)

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')


Christophe  BECAVIN's avatar
Christophe BECAVIN committed
def run_fimo(path_motif, path_script, exp_design_name, motif_list_filename):

    motif_list = list()
    with open(path_motif + motif_list_filename + '.txt', 'r') as motif_file:
        motif_file.readline()
        for row in motif_file:
            motif_list.append(row.split('\t')[0].strip())
Christophe  BECAVIN's avatar
Christophe BECAVIN committed
    data_name = exp_design_name


    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)
Christophe  BECAVIN's avatar
Christophe BECAVIN committed
            fasta_file = path_script + 'fasta/' + data_name + '.fasta'
            background_file = path_script + 'results/' +  data_name + '/background'
            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 + ' --bgfile ' + background_file + ' ' + 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:
        motif_file.readline()
        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_script + 'motif/' + motif + '.meme'
            #print(motif_file)
            for data_name in data_list:
                fasta_file = path_script + 'fasta/' + data_name + '.fasta'
                background_file = path_script + 'results/' +  data_name + '/background'
                folder_centrimo = path_script + 'centrimo/' + motif + '_' + data_name
                centrimo_sh = 'centrimo -seqlen 150 -verbosity 1 -oc ' + folder_centrimo +\
                              ' --bgfile ' + background_file + ' -score 5.0 '\
                              '-ethresh 10.0 ' + fasta_file + ' '+motif_file
                print(centrimo_sh)
                motif_sh.write(centrimo_sh + '\n')


def create_motif_table(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:
        motif_file.readline()
        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()
    fasta_seq = SeqIO.to_dict(SeqIO.parse(path_motif + 'fasta/'+ exp_design_name +'.fasta', "fasta"))
    nb_peaks = len(fasta_seq.keys())
    np_motif = np.zeros((nb_peaks, len(motif_list)))
    df_motif_score = pd.DataFrame(np_motif, index=fasta_seq.keys(), columns=motif_list)
    np_motif = np.zeros((nb_peaks, len(motif_list)))
    df_motif_count = pd.DataFrame(np_motif, index=fasta_seq.keys(), columns=motif_list)
    np_motif = np.ones((nb_peaks, len(motif_list)))
    df_motif_pvalue = pd.DataFrame(np_motif, index=fasta_seq.keys(), columns=motif_list)

    for motif in motif_list:
        print(motif)
        for data in data_list:
            print(data)
            df_motif_peaks = pd.read_csv(path_motif + 'fimo/' + motif + '_' + data + '/fimo.txt', index_col=0, sep='\t')
            grouped = df_motif_peaks[['sequence name','score']].groupby('sequence name')
            score = grouped.sum()
            df_motif_score[motif] = score
            counted = grouped.count()
            df_motif_count[motif] = counted
#           df_motif_pvalue[motif][peak] += pvalue
    df_motif_score.to_csv(path_motif + motif_list_filename + '_score.txt', sep='\t')
    df_motif_count.to_csv(path_motif + motif_list_filename + '_count.txt', sep='\t')

    #df_motif_pvalue.to_csv(path_motif + exp_design_name + '_' + motif_list_filename + '_pvalue.txt', sep='\t')


def motif_vs_fasta(path_motif, path_script, exp_design_name, motif_list_filename, data_list_filename):
    # read fasta summary
    fasta_to_size = dict()
    with open(path_motif + 'Fasta_Summary.txt', 'r') as fasta_summary:
        for row in fasta_summary:
            fasta_name = row.split('\t')[0]
            fasta_size = row.split('\t')[1].strip()
            if fasta_size != '0':
                fasta_to_size[fasta_name] = fasta_size
    # read list
    motif_list = list()
    with open(path_motif + motif_list_filename + '.txt', 'r') as motif_file:
        motif_file.readline()
        for row in motif_file:
            motif_list.append(row.split('\t')[0].strip())
    # read motif summary
    df_motif_score = pd.read_csv(path_motif + motif_list_filename + '_score.txt', index_col=0, sep='\t')
    df_motif_count = pd.read_csv(path_motif + motif_list_filename + '_count.txt', index_col=0, sep='\t')

    # create motif vs fasta tables
    nb_fasta = len(fasta_to_size.keys())
    np_fasta = np.zeros((len(motif_list), nb_fasta))
    df_motif_fasta_score = pd.DataFrame(np_fasta, index=motif_list, columns=fasta_to_size.keys())
    np_fasta = np.zeros((len(motif_list), nb_fasta))
    df_motif_fasta_count = pd.DataFrame(np_fasta, index=motif_list, columns=fasta_to_size.keys())

    print(motif_list)
    # go through all fasta file
    for fasta_name, fasta_size in fasta_to_size.items():
        fasta_peaks = SeqIO.to_dict(SeqIO.parse(path_motif+'fasta/'+fasta_name+'.fasta', 'fasta'))
        #print(fasta_peaks.keys())
        for motif in motif_list:
            motif_score = df_motif_score.loc[fasta_peaks.keys(), motif].sum()
            df_motif_fasta_score[fasta_name][motif] = float(motif_score)/float(fasta_size)
            motif_count = df_motif_count.loc[fasta_peaks.keys(), motif].sum()
            df_motif_fasta_count[fasta_name][motif] = float(motif_count)/float(fasta_size)
    df_motif_fasta_score.to_csv(path_motif + motif_list_filename + '_fasta_score.txt', sep='\t')
    df_motif_fasta_count.to_csv(path_motif + motif_list_filename + '_fasta_count.txt', sep='\t')


#exp_design_name = 'LivOld'
Christophe  BECAVIN's avatar
Christophe BECAVIN committed
exp_design_name = 'LiverZT_MaxValues'
#exp_design_name = 'LivOld_MaxValues'
#exp_design_name = 'CecAm_MaxValues'
#path = '/pasteur/projets/policy01/m6aAkker/'
path = '/Volumes/m6aAkker/'
Christophe  BECAVIN's avatar
Christophe BECAVIN committed
#path_motif = path + 'PeakDiffExpression/' + exp_design_name + '/Motif/'
path_motif = path + 'PeakDiffExpression/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/')
# if not os.path.isdir(path_motif + 'centrimo/'):
#     os.mkdir(path_motif + 'centrimo/')

#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)


Christophe  BECAVIN's avatar
Christophe BECAVIN committed
#motif_list_filename = 'Motif_'+exp_design_name+'_count'
motif_list_filename = 'Motif_List'
# add new motif
# motif = 'RRACH'
Christophe  BECAVIN's avatar
Christophe BECAVIN committed
#motif="NGGACN"
#add_motif(path_motif, motif, motif_list_filename)
#motif = 'NBCAN'
#add_motif(path_motif, motif, motif_list_filename)
# put all motifs figures together to filter the list
Christophe  BECAVIN's avatar
Christophe BECAVIN committed
#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
Christophe  BECAVIN's avatar
Christophe BECAVIN committed
run_fimo(path_motif, path_script, exp_design_name, motif_list_filename)
# write sh file for running centrimo
#run_centrimo(path_motif, path_script, exp_design_name, motif_list_filename, data_list_filename)

# create a table grouping peak marginal score for each motif
#create_motif_table(path_motif, path_script, exp_design_name, motif_list_filename, data_list_filename)
# Count marginal scrore and marginal motif presence for every fasta file
#motif_vs_fasta(path_motif, path_script, exp_design_name, motif_list_filename, data_list_filename)