##############################
#Created on Oct 07 2019

#@author: Yu-Chuan Chang
##############################

##############################
# import libraries
##############################
library(sqldf)
library(RColorBrewer)
library(gplots)
library(amap)

##############################
# set parameters
##############################
str_inputFileName_gtex <- "~/Projects/bmcbioinformatics/GTEx/GTEx_Analysis_v6p_RNA-seq_RNA-SeQCv1.1.8_gene_median_rpkm.gct"
str_outputFilePath <- "~/Projects/bmcbioinformatics/GTEx/"
list_genes <- c("ADAMTS12","APOE","BACH2","CACNA1E","CLEC16A","LINC00299","MICB","PARVA","SYT6","TNKS2","TOB2","VSNL1","CLEC1A","GPR142","IFT20","MFSD6L","PRKAG2-AS1","RPL32")

##############################
# define functions
##############################
customizedHeatmap <- function(df_input){
    matrix_input_heatmap <- data.matrix(df_input[,3:length(df_input)])
    rownames(matrix_input_heatmap) <- df_input[,"Description"]
    #normalization
    matrix_input_heatmap <- (t(apply(matrix_input_heatmap, 1, function(x)scale(x))) + 1) * 50
    colnames(matrix_input_heatmap) <- colnames(df_input)[3:length(df_input)]
    
    #heatmap: clustering function
    fun_h <- function(x) hclust(x, method = "ward.D") # ward.D complete
    fun_Dist <- function(x) Dist(x, method="pearson") # pearson euclidean spearman
    
    #heatmap: plot .png
    color_palette <- colorRampPalette(brewer.pal(9,"PuBu"))(100)
    png(paste0(str_outputFilePath, "./GTEx_Heatmap.png"), height=1200, width=900, pointsize=15)
    heatmap_result <- heatmap.2(matrix_input_heatmap, margins = c(20,7), col=color_palette, distfun=fun_Dist, hclustfun = fun_h, Rowv=T, Colv=F, dendrogram="row", key=T, keysize=1.0, srtCol=90, srtRow=45, key.xlab="Normalized RPKM", labRow=rownames(matrix_input_heatmap), cexRow=1, symkey=F, breaks=c(0:100), trace="none")
    dev.off()
}

##############################
# main function
##############################
df_gtex <- read.table(str_inputFileName_gtex, skip=2, header=T, sep="\t")
df_genelist <- data.frame(GeneSymbol=list_genes)
df_genelist_unique <- sqldf("SELECT DISTINCT(GeneSymbol) FROM df_genelist WHERE GeneSymbol!=''")
df_heatmap <- sqldf("SELECT gtex.* FROM df_genelist_unique gene LEFT JOIN df_gtex gtex ON gene.GeneSymbol==gtex.Description WHERE gtex.Name is not null")
paste("Query", nrow(df_genelist_unique), "genes and", nrow(df_genelist_unique)-nrow(df_heatmap),"of them missed in GTEx Database.", sep=" ")
customizedHeatmap(df_heatmap)