R/plot_impacts.R

In seqCAT: High Throughput Sequencing Cell Authentication Toolkit

#' @title Plot SNV impact distribution
#'
#' @description Plot SNV impact distributions for a binary SNV profile
#'  comparison.
#'
#' @details This function creates publication-ready plots of the impact
#' distribution from a binary dataset comparison across the matched/mismatched
#' SNVs.
#'
#' @export
#' @rdname plot_impacts
#' @param comparison The SNV profile comparison to be plotted.
#' @param legend Show the legend (boolean).
#' @param annotate Annotate each category (boolean).
#' @param annotate_size Text size for annotations (numeric).
#' @param text_size Text size for axes, ticks and legend (numeric).
#' @param palette Colour palette for filling of bars (character vector).
#' @return A ggplot2 graphical object.
#'
#' @examples
#' # Load test comparison data
#' data(test_comparison)
#'
#' # Plot the impact distribution
#' impacts <- plot_impacts(test_comparison)
plot_impacts <- function(comparison,
                         legend        = TRUE,
                         annotate      = TRUE,
                         annotate_size = 9,
                         text_size     = 14,
                         palette       = c("#0D2D59", "#1954A6")) {

    # Matches and impact character vectors
    matches <- c("match", "mismatch")
    impacts <- c("HIGH", "MODERATE", "LOW", "MODIFIER")

    # Subset data to only include overlapping variants
    comparison <- comparison[comparison$match == "match" |
                             comparison$match == "mismatch", ]
    
    # Prioritise multi-impact variants
    comparison <- comparison[c("match", "impact")]
    comparison <- tidyr::separate_(data   = comparison,
                                   col    = "impact",
                                   sep    = ",",
                                   into   = "impact",
                                   extra  = "drop",
                                   remove = TRUE)
    comparison$impact <- gsub("\\[", "", comparison$impact)
    comparison$impact <- gsub("\\]", "", comparison$impact)
    
    # Factorise matches and impacts
    comparison$match <- factor(comparison$match, levels = matches)
    comparison$impact <- factor(comparison$impact, levels = impacts)

    # Calculate impact distribution
    groups <- dplyr::syms(c("match", "impact"))
    count_var <- dplyr::sym("count")
    data <- dplyr::group_by(comparison, !!!groups)
    data <- dplyr::summarise(data, count = dplyr::n())
    data <- dplyr::mutate(data, prop = !!count_var / sum(!!count_var) * 100)
    data <- dplyr::ungroup(data)

    # Add zeroes to empty groups
    for (impact in impacts) {
        for (match in matches) {

            # Check current combination for existing data
            current <- data[data$impact == impact &
                            data$match == match, ]$count

            # Add to non-existing data
            if (length(current) == 0) {

                data[nrow(data) + 1, "match"] <- match
                data[nrow(data), "impact"] <- impact
                data[nrow(data), "count"] <- 0
                data[nrow(data), "prop"] <- 0
            }
        }
    }

    # Create plot object
    gg <- ggplot2::ggplot(data, ggplot2::aes_string(x    = "impact",
                                                    y    = "prop",
                                                    fill = "match")) +
        ggplot2::geom_bar(stat     = "identity",
                          position = "dodge",
                          colour   = "#000000",
                          size     = 0.3) +
        ggplot2::theme_bw() +
        ggplot2::labs(x    = NULL,
                      y    = "Proportion of SNVs in category (%)",
                      fill = NULL) +
        ggplot2::scale_fill_manual(values = palette,
                                   labels = c("Match", "Mismatch")) +
        ggplot2::theme(panel.grid.major.x = ggplot2::element_blank(),
                       text = ggplot2::element_text(size = text_size))

    # Add text annotation (if applicable)
    if (annotate) {

        # Create labels
        data$percent <- paste0(format(data$prop, nsmall = 1,
                                        digits = 1), " %")

        # Add annotations to plot
        gg <- gg +
            ggplot2::geom_text(
                ggplot2::aes_string(label = "count"),
                position = ggplot2::position_dodge(width = 0.9),
                vjust    = -0.5,
                size     = annotate_size / ggplot2::.pt,
                colour   = "#000000") +
            ggplot2::geom_text(
                ggplot2::aes_string(label = "percent"),
                data     = data[data$prop > 5, ],
                position = ggplot2::position_dodge(width = 0.9),
                vjust    = 1.6,
                size     = annotate_size / ggplot2::.pt,
                colour   = "#FFFFFF") +
            ggplot2::geom_text(
                ggplot2::aes_string(label = "percent", y = 0),
                data     = data[data$prop <= 5, ],
                position = ggplot2::position_dodge(width = 0.9),
                vjust    = 1.5,
                size     = annotate_size / ggplot2::.pt,
                colour   = "#4D4D4D")
    }

    # Remove legend (if applicable)
    if (!legend) {
        gg <- gg + ggplot2::theme(legend.position = "none")
    }

    # Return graphics object
    return(gg)
}