###############################################
###  We assume these commands should be     ###
###  performed in the directory which       ###
###  contains six CEL files:                ###
###  GSM189708.CEL, GSM189709.CEL,          ###
###  GSM189710.CEL, GSM189711.CEL,          ###
###  GSM189712.CEL, GSM189713.CEL           ###
###############################################

###############################################
###  1. Preprocessing using six algorithms  ###
###     This outputs six files.             ###
###############################################
library(plier)  # PLIER
library(vsn)    # VSN
library(farms)  # FARMS
library(puma)   # multi-mgMOS (mmgMOS)
library(affy)   # MBEI
library(gcrma)  # GCRMA
                   
data <- ReadAffy()

eset <- justPlier(data) 
write.exprs(eset, file="data_PLIER.txt")
eset <- vsnrma(data) 
write.exprs(eset, file="data_VSN.txt")  
eset <- q.farms(data)
write.exprs(eset, file="data_FARMS.txt")
eset <- mmgmos(data)
write.exprs(eset, file="data_mmgMOS.txt")
eset <- expresso(data, normalize.method="invariantset", bg.correct=FALSE, pmcorrect.method="pmonly", summary.method="liwong")
write.exprs(eset, file="data_MBEI.txt")
eset <- gcrma(data)
write.exprs(eset, file="data_GCRMA.txt")

######################################################
###  2. Analysis using eight gene ranking methods  ###
######################################################

######################################################
###  2-1. Call functions for gene ranking methods  ###
######################################################
WAD <- function(x, cl, dynamic_r, min_v){ 
    x.class1 <- x[(cl == 0)]
    x.class2 <- x[(cl == 1)]
    x_ave <- (mean(x.class1) + mean(x.class2))/2
    weight <- (x_ave - min_v)/dynamic_r
    statistic <- (mean(x.class2) - mean(x.class1))*weight
    return(statistic)
}

AD <- function(x, cl){
    x.class0 <- x[(cl == 0)]
    x.class1 <- x[(cl == 1)]
    statistic <- mean(x.class1) - mean(x.class0)
    return(statistic)
}

FC <- function(x, cl){
    x.class0 <- x[(cl == 0)]
    x.class1 <- x[(cl == 1)]
    statistic <- log2(mean(x.class1)/mean(x.class0))
    return(statistic)
}

library(RankProd)                     # RP
library(st)                           # modT, samT, and shrinkT
source('http://eh3.uc.edu/r/ibmtR.R') # ibmT
library(limma)                        # ibmT
library(ROC)                          # AUC value calculation

######################################
###  2-2. PLIER-preprocessed data  ###
######################################
data <- read.table("data_PLIER.txt", header=TRUE, row.names=1, sep="\t")
data.cl <- c(rep(0, 3), rep(1, 3))
data[data < 0] <- 0

data.degenes <- as.matrix(c(rep(0, nrow(data)))) # True DEGs
rownames(data.degenes) <- rownames(data)         # True DEGs
degenes <- c(                                    # True DEGs
"212154_at",                                     # True DEGs
"212157_at",                                     # True DEGs
"212158_at",                                     # True DEGs
"205207_at",                                     # True DEGs
"209239_at",                                     # True DEGs
"208200_at",                                     # True DEGs
"210118_s_at",                                   # True DEGs
"205798_at",                                     # True DEGs
"210073_at",                                     # True DEGs
"202859_x_at",                                   # True DEGs
"211506_s_at"                                    # True DEGs
)                                                # True DEGs
data.degenes[degenes, ] <- 1                     # True DEGs

tmp0 <- apply(data[,data.cl == 0], 1, mean)               # WAD
tmp1 <- apply(data[,data.cl == 1], 1, mean)               # WAD
tmp_ave <- (tmp0 + tmp1)/2                                # WAD
dr <- max(tmp_ave)-min(tmp_ave)                           # WAD
tmpall <- apply(data, 1, WAD, data.cl, dr, min(tmp_ave))  # WAD
out_WAD <- abs(tmpall)                                    # WAD
rank_WAD <- as.matrix(rank(-out_WAD, ties.method = "min"))# WAD
rownames(rank_WAD) <- rownames(data)                      # WAD
AUC_WAD <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_WAD), rule = dxrule.sca))

out_w <- (tmp_ave - min(tmp_ave))/dr                      # w
rank_w <- as.matrix(rank(-out_w, ties.method = "min"))    # w
rownames(rank_w) <- rownames(data)                        # w
AUC_w <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_w), rule = dxrule.sca))

tmpall <- apply(data, 1, AD, data.cl)                     # AD
out_AD <- abs(tmpall)                                     # AD
rank_AD <- as.matrix(rank(-out_AD, ties.method = "min"))  # AD
rownames(rank_AD) <- rownames(data)                       # AD
AUC_AD <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_AD), rule = dxrule.sca))

data <- 2^data                                            # FC
tmpall <- apply(data, 1, FC, data.cl)                     # FC
out_FC <- abs(tmpall)                                     # FC
data <- log(data, 2)                                      # FC
rank_FC <- as.matrix(rank(-out_FC, ties.method = "min"))  # FC
rownames(rank_FC) <- rownames(data)                       # FC
AUC_FC <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_FC), rule = dxrule.sca))

tmpall <- RP(data, data.cl, num.perm=1, logged=TRUE, na.rm = FALSE, plot = FALSE, rand = 123) # RP
out_RP <- apply(tmpall$RPs, 1, min)                       # RP
rank_RP <- as.matrix(rank(out_RP, ties.method = "min"))   # RP
rownames(rank_RP) <- rownames(data)                       # RP
AUC_RP <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_RP), rule = dxrule.sca))

data.cl <- data.cl + 1
tmpall <- modt.stat(t(data), data.cl)                       # modT
out_modT <- abs(tmpall)                                     # modT
rank_modT <- as.matrix(rank(-out_modT, ties.method = "min"))# modT
rownames(rank_modT) <- rownames(data)                       # modT
AUC_modT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_modT), rule = dxrule.sca))

tmpall <- sam.stat(t(data), data.cl)                        # samT
out_samT <- abs(tmpall)                                     # samT
rank_samT <- as.matrix(rank(-out_samT, ties.method = "min"))# samT
rownames(rank_samT) <- rownames(data)                       # samT
AUC_samT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_samT), rule = dxrule.sca))

tmpall <- shrinkt.stat(t(data), data.cl)                          # shrinkT
out_shrinkT <- abs(tmpall)                                        # shrinkT
rank_shrinkT <- as.matrix(rank(-out_shrinkT, ties.method = "min"))# shrinkT
rownames(rank_shrinkT) <- rownames(data)                          # shrinkT
AUC_shrinkT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_shrinkT), rule = dxrule.sca))

data.cl <- data.cl - 1
design <- model.matrix(~data.cl)                            # ibmT
data <- as.matrix(data)                                     # ibmT
fit <- lmFit(data, design)                                  # ibmT
fit$Amean<-rowMeans(data)                                   # ibmT
fit <- IBMT(fit,2)                                          # ibmT
tmpall <- fit$IBMT.t                                        # ibmT
out_ibmT <- abs(tmpall)                                     # ibmT
rank_ibmT <- as.matrix(rank(-out_ibmT, ties.method = "min"))# ibmT
rownames(rank_ibmT) <- rownames(data)                       # ibmT
AUC_ibmT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_ibmT), rule = dxrule.sca))

AUC_PLIER <- c(AUC_w, AUC_WAD, AUC_AD, AUC_FC, AUC_RP, AUC_modT, AUC_samT, AUC_shrinkT, AUC_ibmT)

######################################
###  2-3. VSN-preprocessed data  ###
######################################
data <- read.table("data_VSN.txt", header=TRUE, row.names=1, sep="\t")
data[data < 0] <- 0

tmp0 <- apply(data[,data.cl == 0], 1, mean)               # WAD
tmp1 <- apply(data[,data.cl == 1], 1, mean)               # WAD
tmp_ave <- (tmp0 + tmp1)/2                                # WAD
dr <- max(tmp_ave)-min(tmp_ave)                           # WAD
tmpall <- apply(data, 1, WAD, data.cl, dr, min(tmp_ave))  # WAD
out_WAD <- abs(tmpall)                                    # WAD
rank_WAD <- as.matrix(rank(-out_WAD, ties.method = "min"))# WAD
rownames(rank_WAD) <- rownames(data)                      # WAD
AUC_WAD <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_WAD), rule = dxrule.sca))

out_w <- (tmp_ave - min(tmp_ave))/dr                      # w
rank_w <- as.matrix(rank(-out_w, ties.method = "min"))    # w
rownames(rank_w) <- rownames(data)                        # w
AUC_w <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_w), rule = dxrule.sca))

tmpall <- apply(data, 1, AD, data.cl)                     # AD
out_AD <- abs(tmpall)                                     # AD
rank_AD <- as.matrix(rank(-out_AD, ties.method = "min"))  # AD
rownames(rank_AD) <- rownames(data)                       # AD
AUC_AD <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_AD), rule = dxrule.sca))

data <- 2^data                                            # FC
tmpall <- apply(data, 1, FC, data.cl)                     # FC
out_FC <- abs(tmpall)                                     # FC
data <- log(data, 2)                                      # FC
rank_FC <- as.matrix(rank(-out_FC, ties.method = "min"))  # FC
rownames(rank_FC) <- rownames(data)                       # FC
AUC_FC <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_FC), rule = dxrule.sca))

tmpall <- RP(data, data.cl, num.perm=1, logged=TRUE, na.rm = FALSE, plot = FALSE, rand = 123) # RP
out_RP <- apply(tmpall$RPs, 1, min)                       # RP
rank_RP <- as.matrix(rank(out_RP, ties.method = "min"))   # RP
rownames(rank_RP) <- rownames(data)                       # RP
AUC_RP <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_RP), rule = dxrule.sca))

data.cl <- data.cl + 1
tmpall <- modt.stat(t(data), data.cl)                       # modT
out_modT <- abs(tmpall)                                     # modT
rank_modT <- as.matrix(rank(-out_modT, ties.method = "min"))# modT
rownames(rank_modT) <- rownames(data)                       # modT
AUC_modT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_modT), rule = dxrule.sca))

tmpall <- sam.stat(t(data), data.cl)                        # samT
out_samT <- abs(tmpall)                                     # samT
rank_samT <- as.matrix(rank(-out_samT, ties.method = "min"))# samT
rownames(rank_samT) <- rownames(data)                       # samT
AUC_samT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_samT), rule = dxrule.sca))

tmpall <- shrinkt.stat(t(data), data.cl)                          # shrinkT
out_shrinkT <- abs(tmpall)                                        # shrinkT
rank_shrinkT <- as.matrix(rank(-out_shrinkT, ties.method = "min"))# shrinkT
rownames(rank_shrinkT) <- rownames(data)                          # shrinkT
AUC_shrinkT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_shrinkT), rule = dxrule.sca))

data.cl <- data.cl - 1
design <- model.matrix(~data.cl)                            # ibmT
data <- as.matrix(data)                                     # ibmT
fit <- lmFit(data, design)                                  # ibmT
fit$Amean<-rowMeans(data)                                   # ibmT
fit <- IBMT(fit,2)                                          # ibmT
tmpall <- fit$IBMT.t                                        # ibmT
out_ibmT <- abs(tmpall)                                     # ibmT
rank_ibmT <- as.matrix(rank(-out_ibmT, ties.method = "min"))# ibmT
rownames(rank_ibmT) <- rownames(data)                       # ibmT
AUC_ibmT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_ibmT), rule = dxrule.sca))

AUC_VSN <- c(AUC_w, AUC_WAD, AUC_AD, AUC_FC, AUC_RP, AUC_modT, AUC_samT, AUC_shrinkT, AUC_ibmT)

######################################
###  2-4. FARMS-preprocessed data  ###
######################################
data <- read.table("data_FARMS.txt", header=TRUE, row.names=1, sep="\t")
data[data < 0] <- 0

tmp0 <- apply(data[,data.cl == 0], 1, mean)               # WAD
tmp1 <- apply(data[,data.cl == 1], 1, mean)               # WAD
tmp_ave <- (tmp0 + tmp1)/2                                # WAD
dr <- max(tmp_ave)-min(tmp_ave)                           # WAD
tmpall <- apply(data, 1, WAD, data.cl, dr, min(tmp_ave))  # WAD
out_WAD <- abs(tmpall)                                    # WAD
rank_WAD <- as.matrix(rank(-out_WAD, ties.method = "min"))# WAD
rownames(rank_WAD) <- rownames(data)                      # WAD
AUC_WAD <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_WAD), rule = dxrule.sca))

out_w <- (tmp_ave - min(tmp_ave))/dr                      # w
rank_w <- as.matrix(rank(-out_w, ties.method = "min"))    # w
rownames(rank_w) <- rownames(data)                        # w
AUC_w <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_w), rule = dxrule.sca))

tmpall <- apply(data, 1, AD, data.cl)                     # AD
out_AD <- abs(tmpall)                                     # AD
rank_AD <- as.matrix(rank(-out_AD, ties.method = "min"))  # AD
rownames(rank_AD) <- rownames(data)                       # AD
AUC_AD <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_AD), rule = dxrule.sca))

data <- 2^data                                            # FC
tmpall <- apply(data, 1, FC, data.cl)                     # FC
out_FC <- abs(tmpall)                                     # FC
data <- log(data, 2)                                      # FC
rank_FC <- as.matrix(rank(-out_FC, ties.method = "min"))  # FC
rownames(rank_FC) <- rownames(data)                       # FC
AUC_FC <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_FC), rule = dxrule.sca))

tmpall <- RP(data, data.cl, num.perm=1, logged=TRUE, na.rm = FALSE, plot = FALSE, rand = 123) # RP
out_RP <- apply(tmpall$RPs, 1, min)                       # RP
rank_RP <- as.matrix(rank(out_RP, ties.method = "min"))   # RP
rownames(rank_RP) <- rownames(data)                       # RP
AUC_RP <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_RP), rule = dxrule.sca))

data.cl <- data.cl + 1
tmpall <- modt.stat(t(data), data.cl)                       # modT
out_modT <- abs(tmpall)                                     # modT
rank_modT <- as.matrix(rank(-out_modT, ties.method = "min"))# modT
rownames(rank_modT) <- rownames(data)                       # modT
AUC_modT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_modT), rule = dxrule.sca))

tmpall <- sam.stat(t(data), data.cl)                        # samT
out_samT <- abs(tmpall)                                     # samT
rank_samT <- as.matrix(rank(-out_samT, ties.method = "min"))# samT
rownames(rank_samT) <- rownames(data)                       # samT
AUC_samT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_samT), rule = dxrule.sca))

tmpall <- shrinkt.stat(t(data), data.cl)                          # shrinkT
out_shrinkT <- abs(tmpall)                                        # shrinkT
rank_shrinkT <- as.matrix(rank(-out_shrinkT, ties.method = "min"))# shrinkT
rownames(rank_shrinkT) <- rownames(data)                          # shrinkT
AUC_shrinkT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_shrinkT), rule = dxrule.sca))

data.cl <- data.cl - 1
design <- model.matrix(~data.cl)                            # ibmT
data <- as.matrix(data)                                     # ibmT
fit <- lmFit(data, design)                                  # ibmT
fit$Amean<-rowMeans(data)                                   # ibmT
fit <- IBMT(fit,2)                                          # ibmT
tmpall <- fit$IBMT.t                                        # ibmT
out_ibmT <- abs(tmpall)                                     # ibmT
rank_ibmT <- as.matrix(rank(-out_ibmT, ties.method = "min"))# ibmT
rownames(rank_ibmT) <- rownames(data)                       # ibmT
AUC_ibmT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_ibmT), rule = dxrule.sca))

AUC_FARMS <- c(AUC_w, AUC_WAD, AUC_AD, AUC_FC, AUC_RP, AUC_modT, AUC_samT, AUC_shrinkT, AUC_ibmT)

######################################
###  2-5. mmgMOS-preprocessed data  ###
######################################
data <- read.table("data_mmgMOS.txt", header=TRUE, row.names=1, sep="\t")
data[data < 0] <- 0

tmp0 <- apply(data[,data.cl == 0], 1, mean)               # WAD
tmp1 <- apply(data[,data.cl == 1], 1, mean)               # WAD
tmp_ave <- (tmp0 + tmp1)/2                                # WAD
dr <- max(tmp_ave)-min(tmp_ave)                           # WAD
tmpall <- apply(data, 1, WAD, data.cl, dr, min(tmp_ave))  # WAD
out_WAD <- abs(tmpall)                                    # WAD
rank_WAD <- as.matrix(rank(-out_WAD, ties.method = "min"))# WAD
rownames(rank_WAD) <- rownames(data)                      # WAD
AUC_WAD <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_WAD), rule = dxrule.sca))

out_w <- (tmp_ave - min(tmp_ave))/dr                      # w
rank_w <- as.matrix(rank(-out_w, ties.method = "min"))    # w
rownames(rank_w) <- rownames(data)                        # w
AUC_w <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_w), rule = dxrule.sca))

tmpall <- apply(data, 1, AD, data.cl)                     # AD
out_AD <- abs(tmpall)                                     # AD
rank_AD <- as.matrix(rank(-out_AD, ties.method = "min"))  # AD
rownames(rank_AD) <- rownames(data)                       # AD
AUC_AD <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_AD), rule = dxrule.sca))

data <- 2^data                                            # FC
tmpall <- apply(data, 1, FC, data.cl)                     # FC
out_FC <- abs(tmpall)                                     # FC
data <- log(data, 2)                                      # FC
rank_FC <- as.matrix(rank(-out_FC, ties.method = "min"))  # FC
rownames(rank_FC) <- rownames(data)                       # FC
AUC_FC <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_FC), rule = dxrule.sca))

tmpall <- RP(data, data.cl, num.perm=1, logged=TRUE, na.rm = FALSE, plot = FALSE, rand = 123) # RP
out_RP <- apply(tmpall$RPs, 1, min)                       # RP
rank_RP <- as.matrix(rank(out_RP, ties.method = "min"))   # RP
rownames(rank_RP) <- rownames(data)                       # RP
AUC_RP <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_RP), rule = dxrule.sca))

data.cl <- data.cl + 1
tmpall <- modt.stat(t(data), data.cl)                       # modT
out_modT <- abs(tmpall)                                     # modT
rank_modT <- as.matrix(rank(-out_modT, ties.method = "min"))# modT
rownames(rank_modT) <- rownames(data)                       # modT
AUC_modT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_modT), rule = dxrule.sca))

tmpall <- sam.stat(t(data), data.cl)                        # samT
out_samT <- abs(tmpall)                                     # samT
rank_samT <- as.matrix(rank(-out_samT, ties.method = "min"))# samT
rownames(rank_samT) <- rownames(data)                       # samT
AUC_samT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_samT), rule = dxrule.sca))

tmpall <- shrinkt.stat(t(data), data.cl)                          # shrinkT
out_shrinkT <- abs(tmpall)                                        # shrinkT
rank_shrinkT <- as.matrix(rank(-out_shrinkT, ties.method = "min"))# shrinkT
rownames(rank_shrinkT) <- rownames(data)                          # shrinkT
AUC_shrinkT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_shrinkT), rule = dxrule.sca))

data.cl <- data.cl - 1
design <- model.matrix(~data.cl)                            # ibmT
data <- as.matrix(data)                                     # ibmT
fit <- lmFit(data, design)                                  # ibmT
fit$Amean<-rowMeans(data)                                   # ibmT
fit <- IBMT(fit,2)                                          # ibmT
tmpall <- fit$IBMT.t                                        # ibmT
out_ibmT <- abs(tmpall)                                     # ibmT
rank_ibmT <- as.matrix(rank(-out_ibmT, ties.method = "min"))# ibmT
rownames(rank_ibmT) <- rownames(data)                       # ibmT
AUC_ibmT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_ibmT), rule = dxrule.sca))

AUC_mmgMOS <- c(AUC_w, AUC_WAD, AUC_AD, AUC_FC, AUC_RP, AUC_modT, AUC_samT, AUC_shrinkT, AUC_ibmT)

######################################
###  2-6. MBEI-preprocessed data  ###
######################################
data <- read.table("data_MBEI.txt", header=TRUE, row.names=1, sep="\t")
data[data < 1] <- 1
data <- log(data, 2)

tmp0 <- apply(data[,data.cl == 0], 1, mean)               # WAD
tmp1 <- apply(data[,data.cl == 1], 1, mean)               # WAD
tmp_ave <- (tmp0 + tmp1)/2                                # WAD
dr <- max(tmp_ave)-min(tmp_ave)                           # WAD
tmpall <- apply(data, 1, WAD, data.cl, dr, min(tmp_ave))  # WAD
out_WAD <- abs(tmpall)                                    # WAD
rank_WAD <- as.matrix(rank(-out_WAD, ties.method = "min"))# WAD
rownames(rank_WAD) <- rownames(data)                      # WAD
AUC_WAD <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_WAD), rule = dxrule.sca))

out_w <- (tmp_ave - min(tmp_ave))/dr                      # w
rank_w <- as.matrix(rank(-out_w, ties.method = "min"))    # w
rownames(rank_w) <- rownames(data)                        # w
AUC_w <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_w), rule = dxrule.sca))

tmpall <- apply(data, 1, AD, data.cl)                     # AD
out_AD <- abs(tmpall)                                     # AD
rank_AD <- as.matrix(rank(-out_AD, ties.method = "min"))  # AD
rownames(rank_AD) <- rownames(data)                       # AD
AUC_AD <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_AD), rule = dxrule.sca))

data <- 2^data                                            # FC
tmpall <- apply(data, 1, FC, data.cl)                     # FC
out_FC <- abs(tmpall)                                     # FC
data <- log(data, 2)                                      # FC
rank_FC <- as.matrix(rank(-out_FC, ties.method = "min"))  # FC
rownames(rank_FC) <- rownames(data)                       # FC
AUC_FC <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_FC), rule = dxrule.sca))

tmpall <- RP(data, data.cl, num.perm=1, logged=TRUE, na.rm = FALSE, plot = FALSE, rand = 123) # RP
out_RP <- apply(tmpall$RPs, 1, min)                       # RP
rank_RP <- as.matrix(rank(out_RP, ties.method = "min"))   # RP
rownames(rank_RP) <- rownames(data)                       # RP
AUC_RP <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_RP), rule = dxrule.sca))

data.cl <- data.cl + 1
tmpall <- modt.stat(t(data), data.cl)                       # modT
out_modT <- abs(tmpall)                                     # modT
rank_modT <- as.matrix(rank(-out_modT, ties.method = "min"))# modT
rownames(rank_modT) <- rownames(data)                       # modT
AUC_modT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_modT), rule = dxrule.sca))

tmpall <- sam.stat(t(data), data.cl)                        # samT
out_samT <- abs(tmpall)                                     # samT
rank_samT <- as.matrix(rank(-out_samT, ties.method = "min"))# samT
rownames(rank_samT) <- rownames(data)                       # samT
AUC_samT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_samT), rule = dxrule.sca))

tmpall <- shrinkt.stat(t(data), data.cl)                          # shrinkT
out_shrinkT <- abs(tmpall)                                        # shrinkT
rank_shrinkT <- as.matrix(rank(-out_shrinkT, ties.method = "min"))# shrinkT
rownames(rank_shrinkT) <- rownames(data)                          # shrinkT
AUC_shrinkT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_shrinkT), rule = dxrule.sca))

data.cl <- data.cl - 1
design <- model.matrix(~data.cl)                            # ibmT
data <- as.matrix(data)                                     # ibmT
fit <- lmFit(data, design)                                  # ibmT
fit$Amean<-rowMeans(data)                                   # ibmT
fit <- IBMT(fit,2)                                          # ibmT
tmpall <- fit$IBMT.t                                        # ibmT
out_ibmT <- abs(tmpall)                                     # ibmT
rank_ibmT <- as.matrix(rank(-out_ibmT, ties.method = "min"))# ibmT
rownames(rank_ibmT) <- rownames(data)                       # ibmT
AUC_ibmT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_ibmT), rule = dxrule.sca))

AUC_MBEI <- c(AUC_w, AUC_WAD, AUC_AD, AUC_FC, AUC_RP, AUC_modT, AUC_samT, AUC_shrinkT, AUC_ibmT)

######################################
###  2-7. GCRMA-preprocessed data  ###
######################################
data <- read.table("data_GCRMA.txt", header=TRUE, row.names=1, sep="\t")
data[data < 0] <- 0

tmp0 <- apply(data[,data.cl == 0], 1, mean)               # WAD
tmp1 <- apply(data[,data.cl == 1], 1, mean)               # WAD
tmp_ave <- (tmp0 + tmp1)/2                                # WAD
dr <- max(tmp_ave)-min(tmp_ave)                           # WAD
tmpall <- apply(data, 1, WAD, data.cl, dr, min(tmp_ave))  # WAD
out_WAD <- abs(tmpall)                                    # WAD
rank_WAD <- as.matrix(rank(-out_WAD, ties.method = "min"))# WAD
rownames(rank_WAD) <- rownames(data)                      # WAD
AUC_WAD <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_WAD), rule = dxrule.sca))

out_w <- (tmp_ave - min(tmp_ave))/dr                      # w
rank_w <- as.matrix(rank(-out_w, ties.method = "min"))    # w
rownames(rank_w) <- rownames(data)                        # w
AUC_w <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_w), rule = dxrule.sca))

tmpall <- apply(data, 1, AD, data.cl)                     # AD
out_AD <- abs(tmpall)                                     # AD
rank_AD <- as.matrix(rank(-out_AD, ties.method = "min"))  # AD
rownames(rank_AD) <- rownames(data)                       # AD
AUC_AD <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_AD), rule = dxrule.sca))

data <- 2^data                                            # FC
tmpall <- apply(data, 1, FC, data.cl)                     # FC
out_FC <- abs(tmpall)                                     # FC
data <- log(data, 2)                                      # FC
rank_FC <- as.matrix(rank(-out_FC, ties.method = "min"))  # FC
rownames(rank_FC) <- rownames(data)                       # FC
AUC_FC <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_FC), rule = dxrule.sca))

tmpall <- RP(data, data.cl, num.perm=1, logged=TRUE, na.rm = FALSE, plot = FALSE, rand = 123) # RP
out_RP <- apply(tmpall$RPs, 1, min)                       # RP
rank_RP <- as.matrix(rank(out_RP, ties.method = "min"))   # RP
rownames(rank_RP) <- rownames(data)                       # RP
AUC_RP <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_RP), rule = dxrule.sca))

data.cl <- data.cl + 1
tmpall <- modt.stat(t(data), data.cl)                       # modT
out_modT <- abs(tmpall)                                     # modT
rank_modT <- as.matrix(rank(-out_modT, ties.method = "min"))# modT
rownames(rank_modT) <- rownames(data)                       # modT
AUC_modT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_modT), rule = dxrule.sca))

tmpall <- sam.stat(t(data), data.cl)                        # samT
out_samT <- abs(tmpall)                                     # samT
rank_samT <- as.matrix(rank(-out_samT, ties.method = "min"))# samT
rownames(rank_samT) <- rownames(data)                       # samT
AUC_samT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_samT), rule = dxrule.sca))

tmpall <- shrinkt.stat(t(data), data.cl)                          # shrinkT
out_shrinkT <- abs(tmpall)                                        # shrinkT
rank_shrinkT <- as.matrix(rank(-out_shrinkT, ties.method = "min"))# shrinkT
rownames(rank_shrinkT) <- rownames(data)                          # shrinkT
AUC_shrinkT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_shrinkT), rule = dxrule.sca))

data.cl <- data.cl - 1
design <- model.matrix(~data.cl)                            # ibmT
data <- as.matrix(data)                                     # ibmT
fit <- lmFit(data, design)                                  # ibmT
fit$Amean<-rowMeans(data)                                   # ibmT
fit <- IBMT(fit,2)                                          # ibmT
tmpall <- fit$IBMT.t                                        # ibmT
out_ibmT <- abs(tmpall)                                     # ibmT
rank_ibmT <- as.matrix(rank(-out_ibmT, ties.method = "min"))# ibmT
rownames(rank_ibmT) <- rownames(data)                       # ibmT
AUC_ibmT <- AUC(rocdemo.sca(truth = as.vector(data.degenes), data = as.vector(-rank_ibmT), rule = dxrule.sca))

AUC_GCRMA <- c(AUC_w, AUC_WAD, AUC_AD, AUC_FC, AUC_RP, AUC_modT, AUC_samT, AUC_shrinkT, AUC_ibmT)

Method <- c("w", "WAD", "AD", "FC", "RP", "modT", "samT", "shrinkT", "ibmT")
tmp <- cbind(Method, AUC_PLIER, AUC_VSN, AUC_FARMS, AUC_mmgMOS, AUC_MBEI, AUC_GCRMA)
write.table(tmp, "result_AUC.txt", sep = "\t", append=F, quote=F, row.names=F)