computeStability <- function(hc1,hc2) {
  stability <- c(1:dim(hc1)[1])
  nbleaves <- c(1:dim(hc1)[1])
  resultat <- .C("computeStability",as.integer(hc1),as.integer(hc2),as.integer(dim(hc1)[1]),as.double(stability),as.integer(nbleaves))
  return(list(resultat[[4]],resultat[[5]]))
}



## AFFICHAGE CLASSE DENDROGRAMME
findInd <- function(h,vectheight) {
  for(i in 1:length(vectheight))
    if(vectheight[i] == h) return(i)
}

decore <- function(dendo,vectheight,stab,minstab,minmembers) {
  if(!is.leaf(dendo)) {
    h <- attr(dendo,"height")
    ind <- findInd(h,vectheight)
    #attr(dendo,"edgetext") <- as.character(round(deco[ind],2))
    if(stab[ind]>=minstab && attr(dendo,"members")>=minmembers)
      {
        attr(dendo,"edgetext") <- as.character(round(stab[ind],2))
        attr(dendo,"edgePar") <- list(p.col="gray99")
      }
    else
      {
        #attr(dendo,"edgetext") <- as.character(round(deco[ind],2))
        #attr(dendo,"edgePar") <- list(p.col="green")
      }        
  }
  dendo
}

decoreDendo <- function(hc,stab,minstab=-Inf,minmembers=0) {
  dendo <- as.dendrogram(hc)
  dendrapply(dendo,decore,hc$height,stab,minstab,minmembers)
}



validTree<-function(array.dat,iterhc=5,interact=F)
  {
    nt<-dim(array.dat)[[2]]
    nrep<-dim(array.dat)[[3]]
    ngene<-dim(array.dat)[[1]]    

#######
#######Matrices  pour strocker les resultats de la simulation
    res<-matrix(0,nrow=iterhc,ncol=ngene-1)
    nbleaves<-rep(0,ngene-1)
    dimnames(res)<-list(paste("iter",1:iterhc,sep=""),
                        paste("nd.",1:(ngene-1),sep=""))
    if(!interact)
      {
        mat<-apply(array.dat,c(1,2),mean)
        matt<-t(apply(mat,1,function(x){x-mean(x)}))
        #att<-mat
        hcall<-hclust(dist(matt),method="ward")
      }
    else
      {
        gamma<-apply(array.dat,c(1,2),mean)
        mtemps<-apply(gamma,2,mean)
        mgene<-apply(gamma,1,mean)
        mu<-mean(gamma)
        gamma<-sweep(gamma,2,mtemps,FUN="-")
        gamma<-sweep(gamma,1,mgene,FUN="-")
        mat<-gamma+mu
        hcall<-hclust(dist(mat),method="ward")
      }

   
    for(i in 1:iterhc)
      {
        cat(paste("Numéro d'itération sur classification =",i,"\r"))
        #rep.cross<-sample(x=c(1:nrep),size=nt,replace=T)
        ar<-array(0,dim=c(ngene,nt,nrep))
        for(j in 1:nt)
          {
            rep.cross<-sample(x=c(1:nrep),size=nrep,replace=T)           
            ar[,j,]<-array.dat[,j,rep.cross]
          }        
    ##On fait la moyenne
        if(!interact)
        {
          matboot<-apply(ar,c(1,2),mean)
          matboott<-t(apply(matboot,1,function(x){x-mean(x)}))
          hcboot<-hclust(dist(matboott),method="ward")
        }
        else
          {
            gamma<-apply(ar,c(1,2),mean)
            mtemps<-apply(gamma,2,mean)
            mgene<-apply(gamma,1,mean)
            mu<-mean(gamma)
            gamma<-sweep(gamma,2,mtemps,FUN="-")
            gamma<-sweep(gamma,1,mgene,FUN="-")
            matboot<-gamma+mu
            hcboot<-hclust(dist(matboot),method="ward")
          }
        tmp <-computeStability(hcall$merge,hcboot$merge)
        res[i,] <- tmp[[1]]
        nbleaves <- tmp[[2]]
      }
    cat("\n")
    invisible(list(stab=res,hc=hcall,nbleaves=nbleaves))
  }





dendnames <- function(n) {
           if(is.leaf(n)) {
            
            lab<- attr(n, "label")
            lab
           }
           
     }