R/utils_mutations.R

In dwassarman/cellpanelr: Analyze Data from Cell Line Panels

#' Number of cell lines present in mutation data set
#'
#' @param data A tibble with "depmap_id" column
#' @param id Column containing DepMap IDs
#'
#' @return integer, number of matching cell lines
n_mut_matched <- function(data, id = "depmap_id") {
  data[[id]] %>%
    intersect(.mut_ids) %>%
    length()
}


#' Generate volcano plot effect vs. p value
#'
#' @param data A tibble, output of cor_mutations()
volcano_plot <- function(data) {
  data %>%
    dplyr::filter(!is.na(.data$log.p)) %>%
    ggplot2::ggplot(ggplot2::aes(
      x = .data$effect,
      y = .data$log.p,
      color = .data$significant
    )) +
    ggplot2::geom_point(alpha = 0.4, size = 4) +
    ggplot2::xlab(bquote(log[2](mutant / wild - type))) +
    ggplot2::ylab(bquote(-log[10](adj.p))) +
    ggplot2::geom_vline(xintercept = 0, linetype = "dashed") +
    ggplot2::geom_hline(yintercept = -log10(0.05), linetype = "dashed")
}


#' Generate boxplot genotype vs. response, faceted by selected gene
#'
#' @param data A tibble, user data merged with mutation data and filtered only for selected genes.
#'   Contains the columns "gene" and "genotype".
#' @param response Name of response column
#' @param log_scale Plot response in log-scale on y axis.
#'
#' @return Boxplot
mut_plot_selected <- function(data, response, log_scale = FALSE) {
  if (nrow(data) == 0) {
    return(NULL)
  }

  p <- data %>%
    ggplot2::ggplot(ggplot2::aes(.data$genotype, .data[[response]])) +
    ggplot2::geom_boxplot(outlier.shape = NA) +
    ggplot2::geom_jitter(ggplot2::aes(color = .data$genotype), width = 0.2, alpha = 0.4) +
    ggplot2::facet_wrap(~ .data$gene) +
    ggplot2::scale_color_viridis_d(option = "C", end = 0.8) +
    ggplot2::theme(legend.position = "none") +
    ggplot2::xlab("Genotype")

  if (log_scale) {
    p <- p + ggplot2::scale_y_log10()
  }

  p
}


#' Tooltip for the mutation volcano plot
#'
#' See here for reference: https://gitlab.com/-/snippets/16220
#'
#' @param data Tibble; output of gene correlation
#' @param hover Hover information from plot
mut_vol_tip <- function(data, hover) {

  # Find point near hover
  point <- nearPoints(
    data,
    hover,
    xvar = "effect",
    yvar = "log.p",
    maxpoints = 1
  )


  # Only show if cursor is near a point
  if (nrow(point) == 0) {
    return(NULL)
  }

  # Get location
  left_px <- hover$coords_css$x
  top_px <- hover$coords_css$y

  # create style property for tooltip
  # background color is set so tooltip is a bit transparent
  # z-index is set so we are sure are tooltip will be on top
  style <- paste0(
    "position:absolute; z-index:100; background-color: rgba(245, 245, 245, 0.85); ",
    "left:", left_px, "px; top:", top_px, "px;"
  )

  # actual tooltip created as wellPanel
  wellPanel(
    style = style,
    strong(point$gene)
  )
}