R/qqunif.R

Defines functions qqunif

Documented in qqunif

#' qqunif
#'
#' Create qqplots with an assumed uniform distribution
#' @param d dataframe with at least one column, p-value; Color, Shape, and Name optional
#' @param CI two-sided confidence interval, default 0.95
#' @param splitby if data contains Color and/or Shape, indicate variable(s) by which the data should be subsetted for calculating CIs
#' @param opacity point opacity, default 1
#' @param title plot title
#' @param groupcolors named vector of colors corresponding to data in Group column
#' @param highlight_name vector of names to highlight, dataframe must include a Name column
#' @param highlight_p p-value threshold to highlight
#' @param highlighter highlighter color
#' @param annotate_name vector of names to annotate, dataframe must include a Name column
#' @param annotate_p p-value threshold to annotate, dataframe must include a Name column
#' @param line draw a red line at pvalue threshold (observed)
#' @param background can change to "white"
#' @param file filename
#' @param wi width of plot
#' @param hgt height of plot
#' @param res resolution of plot
#' @return png image(s)
#' @import ggplot2
#' @export
#' @examples
#' data(gwas)
#' qqdat <- data.frame(pvalue=gwas$pvalue, Color=gwas$Frame)
#' qqunif(d=qqdat, opacity=0.6, splitby="Color")

qqunif <- function(d, CI=0.95, opacity=1, groupcolors, splitby=NULL, highlight_p, highlight_name, annotate_p, annotate_name, highlighter="red", line, background, title, file="qqunif", wi=7, hgt=7, res=300){
    if("Color" %in% colnames(d)){
      if(!missing(groupcolors)){
        colrs <- groupcolors
      } else {
        ngroupcolors <- nlevels(factor(d$Color))
        if(ngroupcolors > 15){
          if (!requireNamespace(c("RColorBrewer"), quietly = TRUE)==TRUE) {
            stop("Please install RColorBrewer to add color attribute for more than 15 colors.", call. = FALSE)
          } else {
            getPalette = grDevices::colorRampPalette(RColorBrewer::brewer.pal(11, "Spectral"))
            colrs<- getPalette(ngroupcolors)
          }
        } else {
          pal <- pal <- c("#009292", "#920000", "#490092", "#db6d00", "#24ff24",
                          "#ffff6d", "#000000", "#006ddb", "#004949","#924900",
                          "#ff6db6", "#6db6ff","#b66dff", "#ffb6db","#b6dbff")
          colrs <- pal[1:ngroupcolors]
        }
      }
    }
    if(!is.null(splitby)){
      dlist <- split(d, d[, splitby])
      df <- lapply(dlist, function(x) cbind(x[order(x$pvalue),],
                                            obs=-log10(sort(x$pvalue)),
                                            ex=-log10(stats::ppoints(length(!is.na(x$pvalue)))),
                                            cl=-log10(stats::qbeta(p = (1-CI)/2, shape1 = 1:length(!is.na(x$pvalue)), shape2 = length(!is.na(x$pvalue)):1)),
                                            cu=-log10(stats::qbeta(p = (1+CI)/2, shape1 = 1:length(!is.na(x$pvalue)), shape2 = length(!is.na(x$pvalue)):1))))
      dat <- do.call("rbind", df)
    } else {
      dat <- cbind(d[order(d$pvalue), , drop=FALSE],
                   obs=-log10(sort(d$pvalue)),
                   ex=-log10(stats::ppoints(length(!is.na(d$pvalue)))),
                   cl=-log10(stats::qbeta(p = (1-CI)/2, shape1 = 1:length(!is.na(d$pvalue)), shape2 = length(!is.na(d$pvalue)):1)),
                   cu=-log10(stats::qbeta(p = (1+CI)/2, shape1 = 1:length(!is.na(d$pvalue)), shape2 = length(!is.na(d$pvalue)):1)))
    }

  if("Shape" %in% splitby & "Color" %in% splitby){
    linaes1 <- "geom_line(aes(ex, cu, linetype=Shape, color=Color))"
    linaes2 <- "geom_line(aes(ex, cl, linetype=Shape, color=Color))"
  } else if("Shape" %in% splitby) {
    linaes1 <- "geom_line(aes(ex, cu, linetype=Shape))"
    linaes2 <- "geom_line(aes(ex, cl, linetype=Shape))"
  } else if("Color" %in% splitby){
    linaes1 <- "geom_line(aes(ex, cu, color=Color))"
    linaes2 <- "geom_line(aes(ex, cl, color=Color))"
  } else {
    linaes1 <- "geom_line(aes(ex, cu), linetype=2)"
    linaes2 <- "geom_line(aes(ex, cl), linetype=2)"
  }
  #Plot
  if("Shape" %in% colnames(d)){
    if("Color" %in% colnames(d)){
      p <- ggplot(dat, aes(ex, obs)) + geom_point(aes(shape=Shape, colour=Color),alpha=opacity)
    } else {
      p <- ggplot(dat, aes(ex, obs)) + geom_point(aes(shape=Shape),alpha=opacity)
    }
    p <- p + theme(legend.position="bottom", legend.title = element_blank())
  } else {
    if("Color" %in% colnames(d)){
      p <- ggplot(dat, aes(ex, obs)) + geom_point(aes(colour=Color), alpha=opacity)
      p <- p + theme(legend.position="bottom", legend.title = element_blank())
    } else{
      p <- ggplot(dat, aes(ex, obs)) + geom_point(alpha=opacity)
    }
  }
  p <- p + eval(parse(text=linaes1)) + eval(parse(text=linaes2))
  p <- p + labs(x=expression(paste("Expected -log"[10], plain(P))), y=expression(paste("Observed -log"[10], plain(P))))
  p <- p + geom_abline(intercept = 0, slope = 1, alpha = 0.5)
  p <- p + theme(panel.grid.minor = element_blank())
  if("Color" %in% colnames(d)){p <- p + scale_color_manual(name = "Color", values = colrs)}
  if(!missing(line)){p <- p + geom_hline(yintercept = redline, colour="red")}
  if(!missing(title)) {p <- p + ggtitle(title)}
  if(!missing(background)) {p <- p + theme(panel.background = element_rect(fill=background))}
  #Add extra annotations
  if(!missing(highlight_name)){
    if("Shape" %in% names(d)){
      p <- p + geom_point(data=dat[dat$Name %in% highlight_name, ], aes(x=ex, y=obs, shape=Shape), colour=highlighter)
      p <- p + guides(shape = guide_legend(override.aes = list(colour = "black")))
    } else {
      p <- p + geom_point(data=dat[dat$Name %in% highlight_name, ], aes(x=ex, y=obs), colour=highlighter)
    }
  }
  if(!missing(highlight_p)){
    if("Shape" %in% names(d)){
      p <- p + geom_point(data=dat[dat$pvalue < highlight_p, ], aes(x=ex, y=obs, shape=Shape), colour=highlighter)
      p <- p + guides(shape = guide_legend(override.aes = list(colour = "black")))
    } else {
      p <- p + geom_point(data=dat[dat$pvalue < highlight_p, ], aes(x=ex, y=obs), colour=highlighter)
    }
  }
  if(!missing(annotate_p)){
    if (!requireNamespace(c("ggrepel"), quietly = TRUE)==TRUE) {
      print("Consider installing 'ggrepel' for improved text annotation")
      p <- p + geom_text(data=dat[dat$pvalue < annotate_p,], aes(ex,obs,label=Name), color="black")
    } else {
      p <- p + ggrepel::geom_text_repel(data=dat[dat$pvalue < annotate_p,], aes(ex,obs,label=Name), color="black")
    }
  }
  if(!missing(annotate_name)){
    if (!requireNamespace(c("ggrepel"), quietly = TRUE)==TRUE){
      print("Consider installing 'ggrepel' for improved text annotation")
      p <- p + geom_text(data=dat[dat$Name %in% annotate_name,], aes(ex,obs,label=Name))
    } else {
      p <- p + ggrepel::geom_text_repel(data=dat[dat$Name %in% annotate_name,], aes(ex,obs,label=Name))
    }
  }
  #Save
  print(paste("Saving plot to ", file, ".png", sep=""))
  ggsave(p, filename=paste(file, ".png", sep=""), dpi=res, units="in", height=hgt, width=wi)
  print(p)

  return(p)

}
anastasia-lucas/plotman documentation built on July 13, 2020, 11:47 p.m.