#####################################
### Processing of methylomic data ###
#####################################

#load the methylomic data from the 450k beadchip
library(minfi)
path<-"G:/Méthylomique/Fichiers IDAT"
list.files(path)
targets<-read.table("G:/Méthylomique/Fichiers IDAT/targets.txt",sep="\t",dec=",",header=TRUE)
rgset<-read.450k.exp(targets=targets)
rgset$Sample_Name<-as.character(rgset$Nom)
rgset$Basename<-as.character(rgset$Basename)

#Preprocessing
MSet<-preprocessRaw(rgset)

#Quality control
qc<-getQC(MSet)
plotQC(qc)
densityPlot(MSet,sampGroups=phenoData$Sample_Group)
densityBeanPlot(MSet, sampGroups = phenoData$Sample_Group)
qcReport(rgset, pdf= "H:/Data AN/qcReport.pdf")

#Correction for cell count
require(FlowSorted.Blood.450k)
cellcounts<-estimateCellCounts(rgset)

#Detection of batch effect
library(sva)
mod=model.matrix(~as.factor(Sample_Group), data=phenoData)
mod #return no batch effect

#Normalization (funnorm)
gset.funnorm<-preprocessFunnorm(rgset)
data<-getBeta(gset.funnorm)

#load the phenotype (converter/non converter)
PHENO<-read.table("phenotype.txt", header=T, sep="\t")

######################################
## Detection of private epimutation ##
######################################


# Compute the variance: Example for the converter group (="C")
list<-subset(PHENO,Statut=="C")
myvars<-list$code.sample
myvars<-as.character(myvars)
newdata<-data[myvars]
Var_Conv<-newdata
Var_Conv$variance<-Var_Conv[,1]
mdata<-data.matrix(newdata)
for (i in 1:nrow(newdata)) {
Var_Conv$variance[i]<-var(mdata[i,])
}
v = cbind(data[,1],Var_Conv$variance)

write.table(v,"variance_converters.txt")

#Redo for non-converter group (="NC")

#F-test for the variance
Clist<-subset(PHENO,Statut=="C")
NClist<-subset(PHENO,Statut=="NC")

myvarsC<-Clist$code.sample
myvarsC<-as.character(myvarsC)

myvarsNC<-NClist$code.sample
myvarsNC<-as.character(myvarsNC)

newdata1<-data[myvarsC]
newdata2<-data[myvarsNC]

nom=as.character(data[,1])
Ftest= matrix(data = NA, nrow = nrow(data), ncol = 1, byrow = FALSE,
       dimnames = list(nom))

for ( i in 1: nrow(newdata1)){
	x=unlist(newdata1[i,])
	y=unlist(newdata2[i,])
	ftest=var.test(x,y,alternative="two.sided")
	Ftest[i]=ftest$p.value
	Ftest$fdr_adjust[i]<-p.adjust(ftest$p.value[i], method = "fdr", n = list(nom))
}

write.table(Ftest,"Ftest_C_vs_NC.txt")


#################################
####### Detection of VMPs #######
#################################

# Compute the median: Example for the converter group (="C")
mymed<-Clist$code.sample
mymed<-as.character(mymed)
newdata<-data[mymed]
Med_Conv<-newdata
Med_Conv$variance<-Med_Conv[,1]
mdata<-data.matrix(newdata)
for (i in 1:nrow(newdata)) {
Med_Conv$median[i]<-median(mdata[i,])
}
m = cbind(data[,1],Med_Conv$median)

write.table(m,"VMP_converters.txt")