Motif exercise script.txt

####EXO Motifs Analysis a partir de Fasta###
source("http://bioconductor.org/biocLite.R")
biocLite("Biostrings")
source("http://bioconductor.org/biocLite.R")
biocLite("seqLogo")
source("http://bioconductor.org/biocLite.R")
biocLite("rGADEM")
source("http://bioconductor.org/biocLite.R")
biocLite("MotIV")

library(rGADEM)
library(MotIV)
library(seqLogo)
library(Biostrings)

####simple sogo from small set####

###EXAMPLE FROM seqLogo###
mFile <- system.file("Exfiles/pwm1", package="seqLogo")
m <- read.table(mFile)
pwm <- makePWM(m)
seqLogo(pwm)
seqLogo(pwm, ic.scale=FALSE)

###EXAMPLE using Biostrings####
patterns <- DNAStringSet(c("GCTAGC", "GGTACC", "GCATGC"))
pwm <- PWM(patterns)
seqLogo(t(t(pwm) * 1/colSums(pwm)))
seqLogo(t(t(pwm) * 1/colSums(pwm)), ic.scale=FALSE)

####EXAMPLE using a fasta and denovo motif finding####
Seq<-read.DNAStringSet("CDG1P.fa", "fasta")
gadem<-GADEM(Seq,verbose=1)
motifs <- getPWM(gadem)
print(motifs)
seqLogo(motifs$m3)

###### comparaison of motif and Jaspar data base of known motifs####
analysis.jaspar <- motifMatch(motifs)
summary(analysis.jaspar)
plot(analysis.jaspar, ncol = 2, top = 10, rev = FALSE, main = "Logo", bysim = TRUE)