server-scripts/qhtcp-workflow/apps/r/createHeatMaps.R

#!/usr/bin/env Rscript
# This script will make heatmaps for the REMc analysis
# need to give the input "finalTable.csv" file after running REMc generated by eclipse

library(RColorBrewer)
library(gplots)

args <- commandArgs(TRUE)

# Set output dir
if (length(args) >= 1) {
  input_finalTable <- file.path(args[1])
} else {
  input_finalTable <- "/REMcHeatmaps/REMcWithShift.csv" # for legacy workflow
}

if (length(args) >= 2) {
  outDir <- file.path(args[2])
} else {
  outDir <- "/REMcHeatmaps/REMcWithShift.csv" # for legacy workflow
}

hmapfile <- data.frame(read.csv(file = input_finalTable, header = TRUE, sep = ",", stringsAsFactors = FALSE))
# set NAs to NA
hmapfile[hmapfile == -100] <- NA
hmapfile[hmapfile == 100] <- NA
hmapfile[hmapfile == 0.001] <- NA
hmapfile[hmapfile == -0.001] <- NA

# select the number of rows based on the number of genes
num_total_genes <- length(hmapfile[, 1])

# Break out the cluster names so each part of the cluster origin can be accessed
# line below removed because it adds to many genes to clusters when going past 1-0-10
# since it cannot differentiate between 1-0-1 and 1-0-10 when using grepl.
# hmapfile$cluster.origin = gsub(" ","",x=hmapfile$cluster.origin)

hmapfile$cluster.origin <- gsub(";", " ;", x = hmapfile$cluster.origin)
hmapfile$cluster.origin <- strsplit(hmapfile$cluster.origin, ";")
#use tail(x,n) for accessing the outward most cluster

clust_rounds <- 0
for (i in 1:num_total_genes) {
  if (length(hmapfile$cluster.origin[[i]]) > clust_rounds) {
    clust_rounds <- length(hmapfile$cluster.origin[[i]])
  }
}

unique_clusts <- unique(hmapfile$cluster.origin[1:num_total_genes])
unique_clusts <- unique_clusts[unique_clusts != " "]

# Select only the unique cluster names
unique_clusts <- sort(unique(unlist(unique_clusts, use.names = FALSE)), decreasing = FALSE)
num_unique_clusts <- length(unique_clusts)

# Base the color key on a statistical analysis of the L and K data
# need to create "breaks" to set the color key, need to have 12 different breaks (for 11 colors)
# scale() will calculate the mean and standard deviation of the entire vector,
# then "scale" each element by those values by subtracting the mean and dividing by the sd.
# hmapfile[,4:(length(hmapfile[1,]) - 2)] <- scale(hmapfile[,4:(length(hmapfile[1,]) - 2)])
# change so that the L data is multiplied to be on the same scale as the K data

KEY_MIN <- 0
KEY_MAX <- 0
K_MIN <- 0
L_MAX <- 0
KcolumnValues <- vector()
LcolumnValues <- vector()

for (i in 4:(length(hmapfile[1, ]) - 2)){
  if (grepl("_Z_lm_K", colnames(hmapfile)[i], fixed = TRUE) == TRUE) {
    KcolumnValues <- append(KcolumnValues, i)
  }
  if (grepl("_Z_lm_L", colnames(hmapfile)[i], fixed = TRUE) == TRUE) {
    LcolumnValues <- append(LcolumnValues, i)
  }
}

# L_MAX <- quantile(hmapfile[,LcolumnValues],c(0,.01,.5,.99,1),na.rm=TRUE)[4]
# K_MIN <- quantile(hmapfile[,KcolumnValues],c(0,.01,.5,.99,1),na.rm=TRUE)[2]
# L_MAX <- quantile(hmapfile[,LcolumnValues],c(0,.01,.5,.975,1),na.rm=TRUE)[4]
# K_MIN <- quantile(hmapfile[,KcolumnValues],c(0,.025,.5,.99,1),na.rm=TRUE)[2]

# Z scores are
L_MAX <- 12
K_MIN <- -12

# L_Multiplier <- as.numeric(abs(K_MIN/L_MAX))
# hmapfile[,LcolumnValues] <- hmapfile[,LcolumnValues] * L_Multiplier

# if(grepl("SHIFT",colnames(hmapfile)[4],fixed=TRUE) == TRUE){
#  print("FOUND SHIFT VALUES")
#  hmapfile[,(LcolumnValues - 1)] <- hmapfile[,(LcolumnValues-1)] * L_Multiplier
# }

# KEY_MAX <- as.numeric(L_MAX * L_Multiplier)
# KEY_MIN <- as.numeric(K_MIN)

KEY_MAX <- as.numeric(L_MAX)
KEY_MIN <- as.numeric(K_MIN)

print(KEY_MIN)
print(L_MAX)
# print(L_Multiplier)

colormapbreaks <- c(KEY_MIN, KEY_MIN * (5 / 6), KEY_MIN * (4 / 6), KEY_MIN * (3 / 6), KEY_MIN * (2 / 6),
  KEY_MIN * (1 / 6), KEY_MAX * (1 / 6), KEY_MAX * (2 / 6), KEY_MAX * (3 / 6), KEY_MAX * (4 / 6), KEY_MAX * (5 / 6), KEY_MAX)
# print(colormapbreaks)

# Probably should give a way to detect shift in case that is is not in the first row... (maybe just grepl for the whole column name?)
# However since also using this to amend the first part.
# Could possibly identify all the ones that contain the word shift and then create an object containing just those numbers
# then could just use these values and create spaces only between interaction values
# possibly could get rid of redundant shift values if we don't want to view these
# could we pool all the shift data/average it?
if (grepl("Shift", colnames(hmapfile)[4], fixed = TRUE) == TRUE) {
  even_columns <- seq(from = 2, to = (length(hmapfile[1, ]) - 7), by = 2)
  # ev_repeat = rep("white",length(even_columns))
  # ev_repeat = rep("red",(length(hmapfile[1,]) - 5))
  # middle_col <- (length(hmapfile[1,]) - 5)/2
  # ev_repeat[(middle_col/2)] <- "black"
  # print(ev_repeat)
}

if (grepl("Shift", colnames(hmapfile)[4], fixed = TRUE) == FALSE) {
  even_columns <- seq(from = 2, to = (length(hmapfile[1, ]) - 7), by = 1)
  print("NO SHIFT VALS FOUND")
}

# FOR THIS SCRIPT ONLY (rap tem hu script)
# even_columns <- c(2,5,7,10,12,15,17)

# m <- 0
colnames_edit <- as.character(colnames(hmapfile)[4:(length(hmapfile[1, ]) - 2)])
# print(colnames_edit)
for (i in 1:length(colnames_edit)) {
  if (grepl("Shift", colnames_edit[i], fixed = TRUE) == TRUE) {
    colnames_edit[i] <- ""
    colnames_edit[i + 1] <- gsub(pattern = "_Z_lm_", replacement = " ", x = colnames_edit[i + 1])
    try(colnames_edit[i + 1] <- gsub(pattern = "_", replacement = " ", x = colnames_edit[i + 1]))
    
    # INT_store <- strsplit(colnames_edit[i+1], "Z_lm")
    # print(length(unlist(INT_store)))
    # if(length(unlist(INT_store)) == 4){
    #   colnames_edit[i+1] <- paste(unlist(INT_store)[1],unlist(INT_store)[2],unlist(INT_store)[3],sep=" ")
    # }
    # if(length(unlist(INT_store)) == 3){
    #
    #   colnames_edit[i+1] <- paste(unlist(INT_store)[1],unlist(INT_store)[2],sep=" ")
    # }
    # if(length(unlist(INT_store)) == 5){
    #   colnames_edit[i+1] <- paste(unlist(INT_store)[1],unlist(INT_store)[2],unlist(INT_store)[3],unlist(INT_store)[4],sep=" ")
    # }
    # if(length(unlist(INT_store)) == 6){
    #   colnames_edit[i+1] <- paste(unlist(INT_store)[1],unlist(INT_store)[2],unlist(INT_store)[6],sep=" ")
    # }
  }
}
print(colnames_edit)
# break()
# colnames_edit[5] <- "TEM HLEG K"
# colnames_edit[10] <- "TEM HL K"
# colnames_edit[15] <- "TEM HLEG L"
# colnames_edit[20] <- "TEM HL L"

# Create the heatmaps
for (i in 1:num_unique_clusts) {
  cluster <- unique_clusts[i]
  cluster_data <- subset(hmapfile, grepl(cluster, cluster.origin))
  cluster_length <- length(cluster_data[, 1])
  if (cluster_length != 1) {
    X0 <- as.matrix(cluster_data[, 4:(length(hmapfile[1, ]) - 2)])
    if (cluster_length >= 2001) {
      mypath <- file.path(outDir, paste0("cluster_", gsub(" ", "", cluster), ".pdf"))
      pdf(file = mypath, height = 20, width = 15)
      heatmap.2(
        x = X0,
        Rowv = TRUE, Colv = NA, distfun = dist, hclustfun = hclust,
        dendrogram = "row", cexCol = 0.8, cexRow = 0.1, scale = "none",
        breaks = colormapbreaks, symbreaks = FALSE, colsep = even_columns, sepcolor = "white", offsetCol = 0.1,
        # zlim=c(-132, 132),
        xlab = "Type of Media", ylab = "Gene Name",
        # cellnote = round(X0,digits = 0), notecex = 0.1, key = TRUE,
        keysize = 0.7, trace = "none", density.info = c("none"), margins = c(10, 8),
        na.color = "red", col = brewer.pal(11, "PuOr"),
        main = cluster,
        # ColSideColors=ev_repeat,
        labRow = as.character(cluster_data$Gene), labCol = colnames_edit
      )
      # abline(v=0.5467,col="black")
      dev.off()
    }
    if (cluster_length >= 201 && cluster_length <= 2000) {
      mypath <- file.path(outDir, paste0("cluster_", gsub(" ", "", cluster), ".pdf"))
      pdf(file = mypath, height = 15, width = 12)
      heatmap.2(
        x = X0,
        Rowv = TRUE, Colv = NA, distfun = dist, hclustfun = hclust,
        dendrogram = "row", cexCol = 0.8, cexRow = 0.1, scale = "none",
        breaks = colormapbreaks, symbreaks = FALSE, colsep = even_columns, sepcolor = "white", offsetCol = 0.1,
        # zlim=c(-132,132),
        xlab = "Type of Media", ylab = "Gene Name",
        cellnote = round(X0, digits = 0), notecex = 0.1, key = TRUE,
        keysize = 0.7, trace = "none", density.info = c("none"), margins = c(10, 8),
        na.color = "red", col = brewer.pal(11, "PuOr"),
        main = cluster,
        labRow = as.character(cluster_data$Gene), labCol = colnames_edit
      )
      # abline(v=0.5316,col="black")
      dev.off()
    }
    if (cluster_length >= 150 && cluster_length <= 200) {
      mypath <- file.path(outDir, paste0("cluster_", gsub(" ", "", cluster), ".pdf"))
      pdf(file = mypath, height = 12, width = 12)
      heatmap.2(
        x = X0,
        Rowv = TRUE, Colv = NA, distfun = dist, hclustfun = hclust,
        dendrogram = "row", cexCol = 0.8, cexRow = 0.1, scale = "none",
        breaks = colormapbreaks, symbreaks = FALSE, colsep = even_columns, sepcolor = "white", offsetCol = 0.1,
        # zlim=c(-132,132),
        xlab = "Type of Media", ylab = "Gene Name",
        cellnote = round(X0, digits = 0), notecex = 0.2, key = TRUE,
        keysize = 1, trace = "none", density.info = c("none"), margins = c(10, 8),
        na.color = "red", col = brewer.pal(11, "PuOr"),
        main = cluster,
        labRow = as.character(cluster_data$Gene), labCol = colnames_edit
      )
      dev.off()
    }
    if (cluster_length >= 101 && cluster_length <= 149) {
      mypath <- file.path(outDir, paste0("cluster_", gsub(" ", "", cluster), ".pdf"))
      pdf(file = mypath, mypath, height = 12, width = 12)
      heatmap.2(
        x = X0,
        Rowv = TRUE, Colv = NA, distfun = dist, hclustfun = hclust,
        dendrogram = "row", cexCol = 0.8, cexRow = 0.2, scale = "none",
        breaks = colormapbreaks, symbreaks = FALSE, colsep = even_columns, sepcolor = "white", offsetCol = 0.1,
        #zlim=c(-132,132),
        xlab = "Type of Media", ylab = "Gene Name",
        cellnote = round(X0, digits = 0), notecex = 0.3, key = TRUE,
        keysize = 1, trace = "none", density.info = c("none"), margins = c(10, 8),
        na.color = "red", col = brewer.pal(11, "PuOr"),
        main = cluster,
        labRow = as.character(cluster_data$Gene), labCol = colnames_edit
      )
      dev.off()
    }
    if (cluster_length >= 60 && cluster_length <= 100) {
      mypath <- file.path(outDir, paste0("cluster_", gsub(" ", "", cluster), ".pdf"))
      pdf(file = mypath, height = 12, width = 12)
      heatmap.2(
        x = X0,
        Rowv = TRUE, Colv = NA, distfun = dist, hclustfun = hclust,
        dendrogram = "row", cexCol = 0.8, cexRow = 0.4, scale = "none",
        breaks = colormapbreaks, symbreaks = FALSE, colsep = even_columns, sepcolor = "white", offsetCol = 0.1,
        # zlim=c(-132,132),
        xlab = "Type of Media", ylab = "Gene Name",
        cellnote = round(X0, digits = 0), notecex = 0.3, key = TRUE,
        keysize = 1, trace = "none", density.info = c("none"), margins = c(10, 8),
        na.color = "red", col = brewer.pal(11, "PuOr"),
        main = cluster,
        labRow = as.character(cluster_data$Gene), labCol = colnames_edit
      )
      dev.off()
    }
    if (cluster_length <= 59 && cluster_length >= 30) {
      mypath <- file.path(outDir, paste0("cluster_", gsub(" ", "", cluster), ".pdf"))
      pdf(file = mypath, height = 9, width = 12)
      heatmap.2(
        x = X0,
        Rowv = TRUE, Colv = NA, distfun = dist, hclustfun = hclust,
        dendrogram = "row", cexCol = 0.8, cexRow = 0.6, scale = "none",
        breaks = colormapbreaks, symbreaks = FALSE, colsep = even_columns, sepcolor = "white", offsetCol = 0.1,
        # zlim=c(-132,132),
        xlab = "Type of Media", ylab = "Gene Name",
        cellnote = round(X0, digits = 0), notecex = 0.4, key = TRUE,
        keysize = 1, trace = "none", density.info = c("none"), margins = c(10, 8),
        na.color = "red", col = brewer.pal(11, "PuOr"),
        main = cluster,
        labRow = as.character(cluster_data$Gene), labCol = colnames_edit
      )
      dev.off()
    }
    if (cluster_length <= 29) {
      mypath <- file.path(outDir, paste0("cluster_", gsub(" ", "", cluster), ".pdf"))
      pdf(file = mypath, height = 7, width = 12)
      heatmap.2(
        x = X0,
        Rowv = TRUE, Colv = NA,
        distfun = dist, hclustfun = hclust,
        dendrogram = "row", cexCol = 0.8, cexRow = 0.9, scale = "none",
        breaks = colormapbreaks, symbreaks = FALSE, colsep = even_columns, sepcolor = "white", offsetCol = 0.1,
        # zlim=c(-132,132),
        xlab = "Type of Media", ylab = "Gene Name",
        cellnote = round(X0, digits = 0), notecex = 0.4, key = TRUE,
        keysize = 1, trace = "none", density.info = c("none"), margins = c(10, 8),
        na.color = "red", col = brewer.pal(11, "PuOr"),
        main = cluster,
        labRow = as.character(cluster_data$Gene), labCol = colnames_edit
      )
      dev.off()
    }
  }
  # print(paste("FINISHED", "CLUSTER",cluster,sep=" "))
}