R/ggUpsetPeaks.R

In amitjavilaventura/seqViewR: Visualization of Omics And Sequencing Data In R

# ggUpsetPeaks() -------------------------

#' @title ggUpsetPeaks
#' @author amitjavilaventura
#'
#' @usage
#' ggUpsetPeaks(peaks, conds = names(peaks), conds_order = cond, order_by_freq = T, num_size = 4, title = NULL, subtitle = NULL, caption = NULL)
#'
#' @seealso plyranges
#'
#' @description
#' Function that calls 'getVennCounts' and draws an ggplot2-based upset plot.
#'
#'
#' @param peaks List of dataframes with the genomic coordinates of the regions to overlap. Dataframes must contain the columns seqnames, start, end.
#' @param conds Character with the same length as the 'peaks' list. The names that are given to the diferente objects in the 'peaks' list. Default: 'names(peaks)'
#' @param conds_order Character of the same length as 'conds'. Same values as in 'conds' but with the desired order of priority. Default: 'conds'.
#' @param num_size Numeric of length 1. Size of the labels indicating the size of each intersection. Default: 4.
#' @param order_by_freq Logical of length 1. If TRUE (the default), the intersections are arranged by descending order of size. Default: TRUE.
#' @param title Character of length 1. Title of the plot. Default: NULL.
#' @param subtitle Character of length 1. Subtitle of the plot. Default: NULL.
#' @param caption Character of length 1. Caption written at the bottom of the plot. Default: NULL.
#'
#' @export

ggUpsetPeaks <- function(peaks, conds = names(peaks), conds_order = conds,
                         order_by_freq = T, num_size = 4,
                         title = NULL, subtitle = NULL, caption = NULL){

  # Load required packages
  require(dplyr)
  require(purrr)
  require(reshape2)
  require(magrittr)
  require(patchwork)
  require(ggplot2)
  require(ggpubr)
  require(ggforestplot)

  # Calculate number of peaks for each condition
  peakNum <- peaks %>% purrr::set_names(conds) %>%
    purrr::imap(~dplyr::mutate(.x, set = .y)) %>%
    bind_rows() %>%
    group_by(set) %>% summarize(n = n()) %>% ungroup()

  # Call makeVenn4upSet to retrieve the count matrix
  x <- getVennCounts(peaks, conds, conds_order, plot = F)

  # Transform count matrix to a data frame
  countMatrix <- x$vennCounts %>%
    matrix(ncol = ncol(x$vennCounts), nrow = nrow(x$vennCounts)) %>%
    as.data.frame() %>%
    set_colnames(colnames(x$vennCounts)) %>%
    set_rownames(rownames(x$vennCounts)) %>%
    filter_all(any_vars(. != 0)) %>%
    tibble::rownames_to_column("Intersection")

  # Arrange by size of intersection
  if(order_by_freq){
    orderInt <- countMatrix %>% dplyr::arrange(desc(Counts)) %>% pull(Intersection)
    countMatrix <- countMatrix %>% dplyr::mutate(Intersection = factor(Intersection, levels = orderInt))
  }

  # Draw the barplot of the intersections
  barPlot <- ggplot(countMatrix, aes(Intersection, Counts)) +
    geom_col(fill = "black", width = 0.5) +
    geom_text(aes(label = Counts, y = Counts + max(Counts)*0.04), size = num_size) +
    scale_y_continuous(expand = c(0,0), limits = c(0, max(countMatrix$Counts)*1.1)) +
    ylab("Intersection size") +
    theme_pubclean(base_size = 10) +
    theme(axis.text.x = element_blank(),
          axis.line.x = element_blank(),
          axis.line.y = element_line(color = "black"),
          axis.ticks.x = element_blank(),
          axis.title.x = element_blank(),
          axis.title.y = element_text(face = "bold"),
          plot.subtitle = element_text(face = "italic"),
          plot.title = element_text(face = "bold"))

  # Draw barplot with size of each set
  setSize <- peakNum %>%
    ggplot(aes(n, set)) +
    geom_col(fill = "black", width = 0.6) +
    ggforestplot::geom_stripes(odd = "#22222222", even = "#00000000") +
    scale_x_reverse(expand = c(0, 0)) +
    scale_y_discrete(position = "right") +
    theme_pubclean(base_size = 10, flip = T) +
    theme(axis.text.y = element_blank(),
          axis.ticks = element_blank(),
          axis.line.x = element_line(color = "black"),
          axis.title.y = element_blank(),
          axis.title.x = element_text(face = "bold"),
          plot.subtitle = element_text(face = "italic"),
          plot.title = element_text(face = "bold")) +
    xlab("Set size")

  # Draw the points for the combination of sets in each intersection
  setPoints <- countMatrix %>%
    dplyr::select(-Counts) %>%
    reshape2::melt() %>%
    filter(value != 0) %>%
    ggplot(aes(y = variable, x = Intersection)) +
    ggforestplot::geom_stripes(odd = "#22222222", even = "#00000000") +
    geom_point(size = 4) +
    geom_line(aes(group = Intersection), size = 1.5) +
    xlab("Intersections") +
    theme_pubclean(base_size = 10) +
    theme(axis.text.x = element_blank(),
          axis.ticks = element_blank(),
          axis.title.y = element_blank(),
          axis.title.x = element_text(face = "bold"),
          plot.subtitle = element_text(face = "italic"),
          plot.title = element_text(face = "bold"),
          axis.line = element_line(color = "black"))


  # Create layout to use with patchwork
  upsetLayout <- "
  ##AAAAA
  ##AAAAA
  ##AAAAA
  ##AAAAA
  BBCCCCC
  "

  # Build upset plot
  upsetPlot <- barPlot + setSize + setPoints +
    plot_layout(design = upsetLayout) +
    plot_annotation(title = title, subtitle = subtitle, caption = caption) &
    theme(plot.title = element_text(face = "bold"),
          plot.subtitle = element_text(face = "italic"))

  # Return plot
  return(upsetPlot)

}