R/barcode_lost.R

In EmanuelSoda/ScreenR: Package to Perform High Throughput Biological Screening

#' @title Count number of barcode lost
#' @description This function counts the number of barcodes lost during the
#'              sequencing. A barcode is lost if its associated shRNA has zero
#'              mapped read in a sample.
#'
#' @param screenR_Object The ScreenR object obtained using the
#'                       \code{\link{create_screenr_object}}
#'
#' @importFrom  dplyr group_by
#' @importFrom  dplyr filter
#' @importFrom  dplyr summarise
#' @importFrom  magrittr %>%
#' @importFrom rlang .data
#' @importFrom dplyr vars
#' @importFrom tidyr drop_na
#' @concept compute
#' @return Return a tibble containing the number of barcode lost for each
#'         sample
#' @examples
#' object <- get0("object", envir = asNamespace("ScreenR"))
#'
#' # In order to count the number of barcodes lost just the ScreenR object is
#' # needed
#' head(barcode_lost(object))
#'
#' @export
barcode_lost <- function(screenR_Object) {
    table <- ScreenR::count_mapped_reads(screenR_Object)
    table <- table %>%
        dplyr::group_by(.data$Sample) %>%
        dplyr::filter(.data$Mapped == 0) %>%
        dplyr::summarise(LostBarcode = n())

    return(table)
}

#' @title Plot number of barcode lost
#' @description This function plots the number of barcode lost in each sample.
#'              Usually lots of barcodes lost mean that the sample has low
#'              quality.
#' @param screenR_Object The ScreenR object obtained using the
#'                       \code{\link{create_screenr_object}}
#' @param palette A vector of colors to be used to fill the barplot.
#' @param alpha A value for the opacity of the plot.
#'              Allowed values are in the range 0 to 1
#' @param legende_position Where to positioning the legend of the plot.
#'                         Allowed values are in the "top", "bottom", "right",
#'                         "left", "none".
#' @importFrom magrittr %>%
#' @importFrom rlang .data
#' @importFrom ggplot2 ggplot aes geom_bar theme geom_text scale_fill_manual
#' @importFrom ggplot2 position_stack
#' @concept plot
#' @return Returns the plot displaying the number of barcode lost in each
#'         sample
#' @examples
#' object <- get0("object", envir = asNamespace("ScreenR"))
#'
#' plot_barcode_lost(object)
#' @export
plot_barcode_lost <- function(screenR_Object, palette = NULL,
    alpha = 1, legende_position = "none") {
    table <- ScreenR::barcode_lost(screenR_Object)

    plot <- ggplot(table, aes(
        x = .data$Sample, y = .data$LostBarcode,
        fill = .data$Sample
    )) +
        geom_bar(stat = "identity", color = "black") +
        theme(legend.position = legende_position) +
        geom_bar(alpha = alpha, stat = "identity", color = "black") +
        geom_text(aes(label = .data$LostBarcode),
            position = position_stack(vjust = 0.8), color = "black",
            size = 5
        )

    if (!is.null(palette)) {
        plot <- plot + scale_fill_manual(values = palette)
    }
    return(plot)
}

#' @title Plot number of barcode lost for  gene
#' @description This function plots the number of barcodes lost in each sample
#'              for each gene. Usually in a genetic screening each gene is
#'              is associated with multiple shRNAs and so barcodes. For this
#'              reason a reasonable number of barcodes associated with the
#'              gene has to be retrieved in order to have a robust result.
#'              Visualizing the number of genes that have lost lot's of barcode
#'              is a Quality Check procedure in order to be aware of the number
#'              of barcode for the hit identified.
#' @param screenR_Object The ScreenR object obtained using the
#'                       \code{\link{create_screenr_object}}
#' @param facet A boolean to use the facet.
#' @param samples A vector of samples that as to be visualize
#' @importFrom magrittr %>%
#' @importFrom rlang .data
#' @importFrom dplyr select_if
#' @importFrom tidyselect all_of
#' @importFrom ggplot2 scale_x_continuous
#' @concept plot
#' @return Return the plot displaying the number of barcode lost for each gene
#'         in each sample.
#' @examples
#' object <- get0("object", envir = asNamespace("ScreenR"))
#'
#' plot_barcode_lost_for_gene(object,
#'     samples = c("Time3_A", "Time3_B")
#' )
#' plot_barcode_lost_for_gene(object,
#'     samples = c("Time3_A", "Time3_B"),
#'     facet = FALSE
#' )
#' @export
plot_barcode_lost_for_gene <- function(screenR_Object, facet = TRUE,
    samples) {
    numericColumn <- screenR_Object@count_table %>%
        dplyr::select_if(is.numeric) %>%
        colnames()

    table <- tidyr::pivot_longer(
        data = screenR_Object@count_table,
        cols = all_of(numericColumn), names_to = "Sample", values_to = "Mapped"
    )
    table <- dplyr::left_join(table, screenR_Object@annotation_table,
        by = "Barcode"
    )
    table <- dplyr::group_by(table, .data$Sample, .data$Gene)

    table <- dplyr::mutate(table, barcode_lost = .data$Mapped == 0)

    table <- dplyr::summarise(table,
        Sample = unique(.data$Sample),
        barcode_lost = sum(.data$barcode_lost), .groups = "drop"
    )
    table <- dplyr::filter(table, .data$barcode_lost != 0)
    table <- tidyr::drop_na(table)
    table <- mutate(table, barcode_lost = as.integer(.data$barcode_lost))

    if (!is.null(samples)) {
        table <- dplyr::filter(table, .data$Sample %in% samples)
    } else {
        stop("The sample vector cannot be NULL.")
    }

    plot <- ggplot(table, aes(.data$barcode_lost, .data$Gene,
        color = .data$Sample
    )) +
        geom_point(show.legend = FALSE) +
        scale_x_continuous(breaks = seq(0, max(table$barcode_lost), 1))

    if (facet) {
        plot <- plot + ggplot2::facet_wrap(vars(.data$Sample), scales = "free")
    }

    return(plot)
}