R/06_Distribution_Plots.R

In PNNL-Comp-Mass-Spec/RomicsProcessor: omics-oriented package for reproducible and sharable data processing pipelines

#' distribBoxplot()
#' @description Plots the boxplot of the sample distribution using ggplot2. The colors used for the plotting will correspond to the main_factor of the romics_object.
#' @param romics_object has to be a log transformed romics_object created using romicsCreateObject() and transformed using the function log2transform() or log10transform()
#' @details create a ggplot2 graphic output displaying the normalized or not intensites (or logged intensities) within each sample. ggplot2 methods can be used to visually adjust the plot.
#' @return a plot generated using ggplot2 is generated with this function.
#' @author Geremy Clair
#' @export
distribBoxplot <- function(romics_object){
  if(!is.romics_object(romics_object) | missing(romics_object)) {stop("romics_object is missing or is not in the appropriate format")}

  data_graph<-reshape2::melt(romics_object$data,id.vars=0)
  fill_graph<-romics_object$colors
  fill_graph<-fill_graph[!is.na(data_graph[,2])]
  data_graph<-data_graph[!is.na(data_graph[,2]),]
  data_graph<-cbind(data_graph,fill_graph)
  colors<-as.character(t(romics_object$metadata[romicsFactorNames(romics_object)=="colors_romics",]))
  factor_graph<-data.frame(romics_object$metadata[romicsFactorNames(romics_object)==romics_object$main_factor,])
  factor_graph<-data.frame(variable=colnames(factor_graph),factor=as.character(t(factor_graph[1,])))
  data_graph<-merge(data_graph,factor_graph,by="variable")


p<-ggplot(aes(x=variable, y=value),data=data_graph)+
  geom_boxplot(aes(fill=factor),alpha=0.8)+
  scale_fill_manual(values=levels(factor(data_graph$fill_graph, levels = unique(data_graph$fill_graph))))+
  ggtitle("Boxplot of the data distribution within each sample")+
  theme_ROP()+
  xlab("Sample")

if(romicsLogCheck(romics_object)==T)
{if(sum(grepl("log2transform\\(",romics_object$steps))>0){
  p<-p+ylab("Log2(intensities)")}else{p<-p+ylab("Log10(intensities)")}}else{p<-p+ylab("Intensities")}

return(p)
}

#' distribViolin()
#' @description Plots the violin plot of the sample distribution using ggplot2. The colors used for the plotting will correspond to the main_factor of the romics_object.
#' @param romics_object has to be a log transformed romics_object created using romicsCreateObject() and transformed using the function log2transform() or log10transform()
#' @details create a ggplot2 graphic output displaying the normalized or not intensites (or logged intensities) within each sample. ggplot2 methods can be used to visually adjust the plot.
#' @return a plot generated using ggplot2 is generated with this function.
#' @author Geremy Clair
#' @export
distribViolin <- function(romics_object){
  if(!is.romics_object(romics_object) | missing(romics_object)) {stop("romics_object is missing or is not in the appropriate format")}

  data_graph<-reshape2::melt(romics_object$data,id.vars=0)
  fill_graph<-romics_object$colors
  fill_graph<-fill_graph[!is.na(data_graph[,2])]
  data_graph<-data_graph[!is.na(data_graph[,2]),]
  data_graph<-cbind(data_graph,fill_graph)
  colors<-as.character(t(romics_object$metadata[romicsFactorNames(romics_object)=="colors_romics",]))
  factor_graph<-data.frame(romics_object$metadata[romicsFactorNames(romics_object)==romics_object$main_factor,])
  factor_graph<-data.frame(variable=colnames(factor_graph),factor=as.character(t(factor_graph[1,])))
  data_graph<-merge(data_graph,factor_graph,by="variable")


  p<-ggplot(aes(x=variable, y=value),data=data_graph)+
    geom_violin(aes(fill=factor),alpha=0.8)+
    scale_fill_manual(values=levels(factor(data_graph$fill_graph, levels = unique(data_graph$fill_graph))))+
    ggtitle("Boxplot of the data distribution within each sample")+
    theme_ROP()+
    xlab("Sample")

  if(romicsLogCheck(romics_object)==T)
  {if(sum(grepl("log2transform\\(",romics_object$steps))>0){
    p<-p+ylab("Log2(intensities)")}else{p<-p+ylab("Log10(intensities)")}}else{p<-p+ylab("Intensities")}

  return(p)
}

#' distribJitter()
#' @description Plots the Jitter plot of the sample distribution using ggplot2. The colors used for the plotting will correspond to the main_factor of the romics_object.
#' @param romics_object has to be a log transformed romics_object created using romicsCreateObject() and transformed using the function log2transform() or log10transform()
#' @details create a ggplot2 graphic output displaying the normalized or not intensites (or logged intensities) within each sample. ggplot2 methods can be used to visually adjust the plot.
#' @return a plot generated using ggplot2 is generated with this function.
#' @author Geremy Clair
#' @export
distribJitter <- function(romics_object){
  if(!is.romics_object(romics_object) | missing(romics_object)) {stop("romics_object is missing or is not in the appropriate format")}

  data_graph<-reshape2::melt(romics_object$data,id.vars=0)
  fill_graph<-romics_object$colors
  fill_graph<-fill_graph[!is.na(data_graph[,2])]
  data_graph<-data_graph[!is.na(data_graph[,2]),]
  data_graph<-cbind(data_graph,fill_graph)
  colors<-as.character(t(romics_object$metadata[romicsFactorNames(romics_object)=="colors_romics",]))
  factor_graph<-data.frame(romics_object$metadata[romicsFactorNames(romics_object)==romics_object$main_factor,])
  factor_graph<-data.frame(variable=colnames(factor_graph),factor=as.character(t(factor_graph[1,])))
  data_graph<-merge(data_graph,factor_graph,by="variable")


  p<-ggplot(aes(x=variable, y=value,fill=factor, colours=fill_graph),data=data_graph)+
    geom_jitter(aes(color=factor),alpha=0.8)+
    scale_color_manual(values=levels(factor(data_graph$fill_graph, levels = unique(data_graph$fill_graph))))+
    ggtitle("Jitter plot of the data distribution within each sample")+
    theme_ROP()+
    xlab("Sample")

  if(romicsLogCheck(romics_object)==T)
  {if(sum(grepl("log2transform\\(",romics_object$steps))>0){
    p<-p+ylab("Log2(intensities)")}else{p<-p+ylab("Log10(intensities)")}}else{p<-p+ylab("Intensities")}

  return(p)
}

#' distribSina()
#' @description Plots the Sina plot of the sample distribution using ggplot2. The colors used for the plotting will correspond to the main_factor of the romics_object.
#' @param romics_object has to be a log transformed romics_object created using romicsCreateObject() and transformed using the function log2transform() or log10transform()
#' @details create a ggplot2 graphic output displaying the normalized or not intensites (or logged intensities) within each sample. ggplot2 methods can be used to visually adjust the plot.
#' @return a plot generated using ggplot2 is generated with this function.
#' @author Geremy Clair
#' @export
distribSina <- function(romics_object){
  if(!is.romics_object(romics_object) | missing(romics_object)) {stop("romics_object is missing or is not in the appropriate format")}

  data_graph<-reshape2::melt(romics_object$data,id.vars=0)
  fill_graph<-romics_object$colors
  fill_graph<-fill_graph[!is.na(data_graph[,2])]
  data_graph<-data_graph[!is.na(data_graph[,2]),]
  data_graph<-cbind(data_graph,fill_graph)
  colors<-as.character(t(romics_object$metadata[romicsFactorNames(romics_object)=="colors_romics",]))
  factor_graph<-data.frame(romics_object$metadata[romicsFactorNames(romics_object)==romics_object$main_factor,])
  factor_graph<-data.frame(variable=colnames(factor_graph),factor=as.character(t(factor_graph[1,])))
  data_graph<-merge(data_graph,factor_graph,by="variable")


  p<-ggplot(aes(x=variable, y=value,fill=factor, colours=fill_graph),data=data_graph)+
    geom_sina(aes(color=factor),alpha=0.8)+
    scale_color_manual(values=levels(factor(data_graph$fill_graph, levels = unique(data_graph$fill_graph))))+
    ggtitle("Sina plot of the data distribution within each sample")+
    theme_ROP()+
    xlab("Sample")

  if(romicsLogCheck(romics_object)==T)
  {if(sum(grepl("log2transform\\(",romics_object$steps))>0){
    p<-p+ylab("Log2(intensities)")}else{p<-p+ylab("Log10(intensities)")}}else{p<-p+ylab("Intensities")}

  return(p)

}

#' distribRidges()
#' @description Plots the ridges plot of the sample distribution using ggplot2. The colors used for the plotting will correspond to the main_factor of the romics_object.
#' @param romics_object has to be a log transformed romics_object created using romicsCreateObject() and transformed using the function log2transform() or log10transform()
#' @details create a ggplot2 graphic output displaying the normalized or not intensites (or logged intensities) within each sample. ggplot2 methods can be used to visually adjust the plot.
#' @return a plot generated using ggplot2 is generated with this function.
#' @author Geremy Clair
#' @export
distribRidges <- function(romics_object){
  if(!is.romics_object(romics_object) | missing(romics_object)) {stop("romics_object is missing or is not in the appropriate format")}

  data_graph<-reshape2::melt(romics_object$data,id.vars=0)
  fill_graph<-romics_object$colors
  fill_graph<-fill_graph[!is.na(data_graph[,2])]
  data_graph<-data_graph[!is.na(data_graph[,2]),]
  data_graph<-cbind(data_graph,fill_graph)
  colors<-as.character(t(romics_object$metadata[romicsFactorNames(romics_object)=="colors_romics",]))
  factor_graph<-data.frame(romics_object$metadata[romicsFactorNames(romics_object)==romics_object$main_factor,])
  factor_graph<-data.frame(variable=colnames(factor_graph),factor=as.character(t(factor_graph[1,])))
  data_graph<-merge(data_graph,factor_graph,by="variable")


  p<-ggplot(aes(x=value, y=variable),data=data_graph)+
    geom_density_ridges(scale=1.5,aes(fill=factor),alpha=0.8,panel_scaling = T)+
    scale_fill_manual(values=levels(factor(data_graph$fill_graph, levels = unique(data_graph$fill_graph))))+
    ggtitle("Density ridges plot of the data distribution within each sample")+
    theme_ROP()+
    ylab("Sample")

  if(romicsLogCheck(romics_object)==T)
  {if(sum(grepl("log2transform\\(",romics_object$steps))>0){
    p<-p+xlab("Log2(intensities)")}else{p<-p+xlab("Log10(intensities)")}}else{p<-p+xlab("Intensities")}

  return(p)
}

#' distribHistogram()
#' @description Plots individual frequency plots showing the discribution of the data in each sample using grid.Extra and ggplots2. Please use the function distribHistogramGlobal() to plot the global distribution of the whole dataset.
#' @param romics_object has to be a log transformed romics_object created using romicsCreateObject() and transformed using the function log2transform() or log10transform()
#' @param bin has to be a numerical value indicating the width of the frequency bins to use for the visualization
#' @param scale_y has to be a numerical/double vector of length 2 indicating the y_limits of each graph. if too low those will be automatically adjusted. by default  scale_y=c(0,100) will be used
#' @param scale_x has to be a numerical/double vector of length 2 indicating the x_limits of each graph. if too low those will be automatically adjusted. by default  scale_x=c(0,100) will be used
#' @param col has to be a double vector of lenght 1 indicating in how many columns the graphics will be displayed.
#' @details plots a complex graphic output displaying the data distribution within each sample
#' @return a plot generated using ggplot2 is generated with this function.
#' @author Geremy Clair
#' @export
distribHistogram<- function(romics_object, bin=1, scale_y=c(0,100), scale_x=c(-10,10), col=3){
  if(!is.romics_object(romics_object) | missing(romics_object)) {stop("romics_object is missing or is not in the appropriate format")}
  if(missing(bin)){bin<-1}
  if(missing(col)){col<-3}
  if(!missing(scale_x) & !(is.numeric(scale_x) | is.double(scale_x)) & length(scale_x)!=2) {stop("scale_x has to be a numerical or double vector of lenght 2")}
  if(!missing(scale_y) & !(is.numeric(scale_y) | is.double(scale_y)) & length(scale_y)!=2) {stop("scale_y has to be a numerical or double vector of lenght 2")}
  if(missing(scale_x)){scale_x=c(-10,10)}
  if(missing(scale_y)){scale_y=c(0,100)}

  data<-as.matrix(romics_object$data)
  fill<-as.character(t(romics_object$metadata[grepl("colors_romics",rownames(romics_object$metadata)),]))

  #Check X bins and modify accordingly

  if(!missing(scale_x) & scale_x[2]< max(data, na.rm = TRUE)){
    scale_x[2]<-ceiling(max(data, na.rm = TRUE))
    warning("Your scale_x maximum value was too low, it was adjusted")
  }
  if(!missing(scale_x) & scale_x[1]> min(data, na.rm = TRUE)){
    scale_x[1]<- floor(min(data, na.rm = TRUE))
    warning("Your scale_x minimum value was too high, it was adjusted")
  }

  #check Y bins and modify accodingly (only top value)
  #estimate the break histogram
  breaks<- seq(scale_x[1],scale_x[2],by=bin)
  #create frequency tables

  freq_table <-data.frame(hist(data[!is.na(data[1]),1],breaks,plot = FALSE)$counts)
  for (i in 2:ncol(data)){
    freq_table<- cbind(freq_table,hist(data[!is.na(data[i]),i],breaks,plot = FALSE)$counts)
  }

  colnames(freq_table)<- colnames(data)
  freq_table$bins<-breaks[1:nrow(freq_table)]

  if(!missing(scale_y) & scale_y[1]> 0){
    warning("Your scale_y minimum was higher than 0, it was set at 0")
    scale_y[1]<-0
  }

  if(!missing(scale_y) & scale_y[2]<max(freq_table)){
    scale_y[2]<- max(freq_table)
    warning("Your scale_y maximum was too low, it was adjusted to allow to visualize all the data")
  }

  #generate the plots
  myplots <- list()
  for (i in 1:ncol(data))
    local({
      i <- i
      p <- ggplot()+aes(data[,i])+
        geom_histogram(binwidth=bin, fill=fill[i],alpha=I(.8))+
        ggtitle(names(data)[i])+theme_ROP()+
        scale_y_continuous(limits=scale_y)+
        scale_x_continuous(limits=scale_x)+
        ylab("Frequency")+
        ggtitle(colnames(romics_object$data)[i])

        if(romicsLogCheck(romics_object)==T)
        {if(sum(grepl("log2transform\\(",romics_object$steps))>0){
          p<-p+xlab("Log2(intensities)")}else{p<-p+xlab("Log10(intensities)")}}else{p<-p+xlab("Intensities")}

        myplots[[i]] <<- p
    })
  #generate the text of the plotting function
  plot_hist<-"grid.arrange("
  for (i in 1:(length(myplots))){
    plot_hist<-paste0(plot_hist,"myplots[[",i,"]],")
  }
  plot_hist<-paste0(plot_hist,"ncol=",col,")")

  #run the script stored in plot_hist
  options(warn = -1)
  p<-eval(parse(text = plot_hist))
  options(warn = 1)

}

#' distribHistogramGlobal()
#' @description Plots the frequency plot showing the discribution of the data in the whole data layer using ggplots2. Please use the function distribHistogram() to plot the global distribution for each sample.
#' @param romics_object has to be a log transformed romics_object created using romicsCreateObject() and transformed using the function log2transform() or log10transform()
#' @param bin has to be a numerical value indicating the width of the frequency bins to use for the visualization
#' @details plots a complex graphic output displaying the data distribution within each sample
#' @return a plot generated using ggplot2 is generated with this function.
#' @author Geremy Clair
#' @export
distribHistogramGlobal<-function(romics_object,bin=1){
  if(missing(romics_object)){stop("romics_object is missing")}
  if(class(romics_object)!="romics_object"){stop("your romics_object was not created using the function romicsCreateObject")}
  if(romics_object$steps[1]!="romics_object"){stop("romics_object is not in the appropriate format")}
  if(missing(bin)){bin=1}

  melted_data<-reshape2::melt(romics_object$data)
  info_data<-summary(melted_data$value)

  p<- ggplot(melted_data, aes(x = value),fill="gray") +
    geom_histogram(position="identity", alpha=0.8,binwidth = bin)+
    xlab("data distribution")+
    theme_ROP()+
    scale_fill_manual(values=c("gray"))+
    ylab("Frequency")+
    ggtitle("Distribution Frequency of the data")

  if(romicsLogCheck(romics_object)==T)
  {if(sum(grepl("log2transform\\(",romics_object$steps))>0){
    p<-p+xlab("Log2(intensities)")}else{p<-p+xlab("Log10(intensities)")}}else{p<-p+xlab("Intensities")}

  return(p)
  }