Add local files for motif analysis

motif.sh

+
+project=$1
+comparison=$2
+PATH=/Users/cbecavin/Documents/m6aAkker/PeakDiffExpression/$project
+
+PATH_MOTIF=$PATH'/Motif/results/'$comparison
+echo $PATH_MOTIF
+#'mkdir '$PATH_MOTIF
+
+project=CecAm
+PATH_MOTIF=/pasteur/projets/policy01/m6aAkker/PeakDiffExpression/$project/Motif
+PATH_RESULT=${PATH_MOTIF}/results/All_$project'_NEG'
+PATH_NEG_fasta=$PATH_MOTIF/fasta/All_$project'_NoMeth.fasta'
+PATH_fasta=$PATH_MOTIF/fasta/All_$project'_Meth.fasta'
+
+meme-chip -oc $PATH_RESULT -time 300 -neg $PATH_NEG_fasta -index-name meme-chip.html -order 1 -meme-mod anr -meme-minw 4 -meme-maxw 7 -meme-nmotifs 3 -dreme-e 0.05 -centrimo-score 5.0 -centrimo-ethresh 10.0 $PATH_fasta
+
+project=CecAm
+PATH_MOTIF=/pasteur/projets/policy01/m6aAkker/PeakDiffExpression/$project/Motif
+PATH_RESULT=${PATH_MOTIF}/results/Max_$project'_NEG'
+PATH_NEG_fasta=$PATH_MOTIF/fasta/Max_$project'_NoMeth.fasta'
+PATH_fasta=$PATH_MOTIF/fasta/Max_$project'_Meth.fasta'
+
+meme-chip -oc $PATH_RESULT -time 300 -neg $PATH_NEG_fasta -index-name meme-chip.html -order 1 -meme-mod anr -meme-minw 4 -meme-maxw 7 -meme-nmotifs 3 -dreme-e 0.05 -centrimo-score 5.0 -centrimo-ethresh 10.0 $PATH_fasta
+
+project=LivOld
+PATH_MOTIF=/pasteur/projets/policy01/m6aAkker/PeakDiffExpression/$project/Motif
+PATH_RESULT=${PATH_MOTIF}/results/All_$project'_NEG'
+PATH_NEG_fasta=$PATH_MOTIF/fasta/All_$project'_NoMeth.fasta'
+PATH_fasta=$PATH_MOTIF/fasta/All_$project'_Meth.fasta'
+
+meme-chip -oc $PATH_RESULT -time 300 -neg $PATH_NEG_fasta -index-name meme-chip.html -order 1 -meme-mod anr -meme-minw 4 -meme-maxw 7 -meme-nmotifs 3 -dreme-e 0.05 -centrimo-score 5.0 -centrimo-ethresh 10.0 $PATH_fasta
+
+project=LivOld
+PATH_MOTIF=/pasteur/projets/policy01/m6aAkker/PeakDiffExpression/$project/Motif
+PATH_RESULT=${PATH_MOTIF}/results/Max_$project'_NEG'
+PATH_NEG_fasta=$PATH_MOTIF/fasta/Max_$project'_NoMeth.fasta'
+PATH_fasta=$PATH_MOTIF/fasta/Max_$project'_Meth.fasta'
+
+meme-chip -oc $PATH_RESULT -time 300 -neg $PATH_NEG_fasta -index-name meme-chip.html -order 1 -meme-mod anr -meme-minw 4 -meme-maxw 7 -meme-nmotifs 3 -dreme-e 0.05 -centrimo-score 5.0 -centrimo-ethresh 10.0 $PATH_fasta
+
+
+
+
+PATH_fasta=$PATH'/Motif/fasta/'$comparison'.fasta'
+meme_command='meme-chip -oc '$PATH_MOTIF' -index-name meme-chip.html -time 300 -order 1 -meme-mod zoops -meme-minw 4 -meme-maxw 7 -meme-nmotifs 3 -dreme-e 0.05 -centrimo-score 5.0 -centrimo-ethresh 10.0 '$PATH_fasta
+echo $meme_command >> '/Users/cbecavin/Documents/m6aAkker/Scripts/all_motif_'$project'.sh'
+
+folderDreme=/Users/cbecavin/Documents/m6aAkker/PeakDiffExpression/CecAm/Motif/results/Max_CecAm
+folderFimo=/Users/cbecavin/Documents/m6aAkker/PeakDiffExpression/CecAm/Motif/fimo/All_CecAm
+folderCentr=/Users/cbecavin/Documents/m6aAkker/PeakDiffExpression/CecAm/Motif/centrimo/Max_CecAm
+
+
+fimo --parse-genomic-coord --verbosity 1 --no-qvalue --thresh 5e-1 --max-stored-scores 1000000 --oc $folderFimo --bgfile $folderDreme/background $folderDreme/../../m6a_motif.meme $folderDreme/All_CecAm.fasta
+fimo --parse-genomic-coord --verbosity 1 --thresh 1e-2 --max-stored-scores 1000000 --oc $folderFimo $folderDreme/../../m6a_motif.meme $folderDreme/All_CecAm.fasta
+
+fimo --parse-genomic-coord --verbosity 1 --thresh 5e-1 --oc $folderFimo $folderDreme/../../m6a_motif.meme $folderDreme/All_CecAm.fasta
+fimo --parse-genomic-coord --verbosity 1 --thresh 5e-1 --max-stored-scores 1000000 --oc $folderFimo $folderDreme/../../m6a_motif.meme $folderDreme/All_CecAm.fasta
+
+
+centrimo -seqlen 100 -verbosity 1 -oc $folderCentr -score 5.0 -ethresh 10.0 $folderDreme/Max_CecAm.fasta $folderDreme/../../m6a_motif.meme
\ No newline at end of file

motif.txt

+fasta-get-markov -nostatus -m 1 < /pasteur/projets/policy01/m6aAkker/PeakDiffExpression/CecAm/Motif/results/All_CecAm/All_CecAm.fasta 1> /pasteur/projets/policy01/m6aAkker/PeakDiffExpression/CecAm/Motif/results/All_CecAm/background
+
+  getsize /pasteur/projets/policy01/m6aAkker/PeakDiffExpression/CecAm/Motif/results/All_CecAm/All_CecAm.fasta 1> $metrics
+
+  fasta-most -min 50 < /pasteur/projets/policy01/m6aAkker/PeakDiffExpression/CecAm/Motif/results/All_CecAm/All_CecAm.fasta 1> $metrics
+
+  fasta-center -len 100 < /pasteur/projets/policy01/m6aAkker/PeakDiffExpression/CecAm/Motif/results/All_CecAm/All_CecAm.fasta 1> /pasteur/projets/policy01/m6aAkker/PeakDiffExpression/CecAm/Motif/results/All_CecAm/seqs-centered
+
+  fasta-dinucleotide-shuffle -f /pasteur/projets/policy01/m6aAkker/PeakDiffExpression/CecAm/Motif/results/All_CecAm/seqs-centered -t -dinuc 1> /pasteur/projets/policy01/m6aAkker/PeakDiffExpression/CecAm/Motif/results/All_CecAm/seqs-shuffled
+
+
+  dreme -v 1 -oc /pasteur/projets/policy01/m6aAkker/PeakDiffExpression/CecAm/Motif/results/All_CecAm/dreme_out -p /pasteur/projets/policy01/m6aAkker/PeakDiffExpression/CecAm/Motif/results/All_CecAm/seqs-centered -n /pasteur/projets/policy01/m6aAkker/PeakDiffExpression/CecAm/Motif/results/All_CecAm/seqs-shuffled -png -t 6781 -e 0.05
+
+  fimo --parse-genomic-coord --verbosity 1 --oc /pasteur/projets/policy01/m6aAkker/PeakDiffExpression/CecAm/Motif/results/All_CecAm/fimo_out_2 --bgfile /pasteur/projets/policy01/m6aAkker/PeakDiffExpression/CecAm/Motif/results/All_CecAm/background --motif AKRAA /pasteur/projets/policy01/m6aAkker/PeakDiffExpression/CecAm/Motif/results/All_CecAm/dreme_out/dreme.xml /pasteur/projets/policy01/m6aAkker/PeakDiffExpression/CecAm/Motif/results/All_CecAm/All_CecAm.fasta
+
+  centrimo -seqlen 401 -verbosity 1 -oc /pasteur/projets/policy01/m6aAkker/PeakDiffExpression/CecAm/Motif/results/All_CecAm_Meth/centrimo_out -bgfile /pasteur/projets/policy01/m6aAkker/PeakDiffExpression/CecAm/Motif/results/All_CecAm_Meth/background -score 5.0 -ethresh 10.0 /pasteur/projets/policy01/m6aAkker/PeakDiffExpression/CecAm/Motif/results/All_CecAm_Meth/All_CecAm_Meth.fasta /pasteur/projets/policy01/m6aAkker/PeakDiffExpression/CecAm/Motif/results/All_CecAm_Meth/meme_out/meme.xml /pasteur/projets/policy01/m6aAkker/PeakDiffExpression/CecAm/Motif/results/All_CecAm_Meth/dreme_out/dreme.xml 
\ No newline at end of file

src/motif_analysis.R

+library(pheatmap)
+library(GGally)
+library(entropy)
+library(ExPosition)
+library(outliers)
+
+general_path <- "/Users/cbecavin/Documents/m6aAkker/"
+#general_path <- "/pasteur/projets/policy01/m6aAkker/"
+
+project_name = "CecAm"
+#name_setup = "LivOld"
+
+project_dir <- paste(general_path,"PeakDiffExpression/",project_name,"/Motif",sep="")
+setwd(project_dir)
+
+
+tab <- read.table("Fimo_Table_Max.txt", header=TRUE, sep="\t", row.names = 1)
+hm <- pheatmap(log10(tab), scale = "column")
+g <- factor(cutree(hm$tree_row, 3))
+ggpairs(data.frame(log10(tab), g), columns = 1:6, 
+        ggplot2::aes(colour=g))
+
+mi.empirical(tab)
+mi.shrink(tab)
+
+# 50 bins, sur la variable X_meth
+mutinf <- function(vec) {
+  xfac <- cut(vec, breaks = 10)
+  xnominal <-makeNominalData(data.frame(xfac))
+  mi.empirical(xnominal)
+}
+
+mutinf(setNames(log10(tab$X), rownames(tab)))
+mutinf(setNames(log10(tab$X_MethNoGene), rownames(tab)))
+mutinf(setNames(log10(tab$X_Meth_Epi), rownames(tab)))
+mutinf(setNames(log10(tab$X_Meth), rownames(tab)))
+
+# test d'outlier
+library(outliers)
+
+outlier(log10(tab$X))