R/visualisation.R

In RausellLab/CelliD: Unbiased Extraction of Single Cell gene signatures using Multiple Correspondence Analysis

#' Seurat DimPlot for MCA like Dimensionality Reduction
#'
#' @description Small modification of the regular Seurat DimPlot function to enable plotting features for mca like dimensionality reduction.
#' Allows to represent a set of genes of interest on top of the regular cell scatter plot. 
#' The label of the genes can be iverlayed also but it is recommended to plot less than 50 genes label as it can overcrowd the plot severely.     
#'
#' @param X a Seurat object
#' @param reduction Which dimensionality reduction to use. If not specified, searches for mca.
#' @param dims Dimensions to plot, must be a two-length numeric vector specifying x- and y-dimensions
#' @param features character vector of features to plot, must be present in the specified dimension loadings
#' @param size.feature integer indicating size of geom_point for features
#' @param size.feature.text integer indicating size of geom_text for features
#' @param as.text logical indicating as to include text label for feature plotting, will produce warning if TRUE and length(features) > 50
#' @param ... Other arguments passed to DimPlot
#'
#' @importFrom Seurat DimPlot
#' @return A ggplot object
#' @export
#'
#' @examples
#' seuratPbmc <- RunMCA(seuratPbmc, nmcs = 5)
#' seuratPbmc <- DimPlotMC(seuratPbmc, features = Seurat::VariableFeatures(seuratPbmc))
DimPlotMC <-
    function(X, reduction = "mca", dims = c(1, 2), features = NULL, size.feature = 2, size.feature.text = 5, as.text = FALSE, ...) {
        check <- checkCelliDArg(
            X = X,
            dims = dims,
            reduction = reduction,
            features = features
        )
        features <- check$features
        dims <- check$dims
        CellData <- as.data.table(Embeddings(X, reduction)[, dims])
        featureData <- as.data.table(Loadings(X, reduction)[features, dims])
        if (length(features) == 1) {
            featureData <- as.data.frame(t(featureData))
        }
        featureData$features <- features
        featureData$genes <- "black"
        DIM1 <- colnames(featureData)[1]
        DIM2 <- colnames(featureData)[2]
        MCPlot <-
            DimPlot(X, dims = dims, reduction = reduction, ...) + geom_point(
                data = featureData,
                mapping = aes_string(x = DIM1, y = DIM2, text = "features", fill = "genes"),
                size = size.feature,
                shape = 4
            ) + scale_fill_identity(name = "genes", labels = c(""), guide = "legend")
        if (as.text) {
            requireNamespace("ggrepel", quietly = TRUE)
            if (length(features) > 50) {
                if (menu(
                    choices = c("Yes", "No"),
                    title = "you are plotting as Text more than 50 genes, this can take a long time to plot, continue?"
                ) == 2) {
                    stop()
                }
            }
            MCPlot <-
                MCPlot + ggrepel::geom_text_repel(
                    data = featureData,
                    mapping = aes_string(
                        x = DIM1,
                        y = DIM2,
                        text = "features",
                        label = "features"
                    ),
                    size = size.feature.text,
                    seed = 1,
                    min.segment.length = 0.01,
                    point.padding = 0.2
                )
        }
        return(MCPlot)
    }


#' Scater plotReducedDim for MCA like dimensionality Reduction
#'
#' @description Small modification of the Scater plotReducedDim function to enable plotting features for mca like dimensionality reduction.
#' Allows to represent a set of genes of interest on top of the regular cell scatter plot. 
#' The label of the genes can be iverlayed also but it is recommended to plot less than 50 genes label as it can overcrowd the plot severely.    
#' 
#' @param X a Single Cell Experiment Object
#' @param reduction Which dimensionality reduction to use. If not specified, searches for mca.
#' @param dims Dimensions to plot, must be a two-length numeric vector specifying x- and y-dimensions
#' @param features character vector of features to plot, must be present in the specified dimension loadings
#' @param size.feature integer indicating size of geom_point for features
#' @param size.feature.text integer indicating size of geom_text for features
#' @param as.text logical indicating as to include text label for feature plotting, will produce warning if TRUE and length(features) > 50.
#' @param ... Other arguments passed to plotReducedDim
#'
#' @return A ggplot object
#'
#' @export
#' @importFrom scater plotReducedDim
#'
#' @examples
#' scePBMC <- as.SingleCellExperiment(seuratPbmc)
#' scePBMC <- RunMCA(scePBMC, nmcs = 5)
#' plotReducedDimMC(scePBMC)
plotReducedDimMC <-
    function(X, reduction = "MCA", dims = c(1, 2), features = NULL, size.feature = 3, size.feature.text = 5, as.text = FALSE, ...) {
        check <- checkCelliDArg(X, reduction = reduction, dims = dims, features = features, group.by = NULL)
        features <- check$features
        dims <- check$dims
        featureData <- as.data.frame(attr(reducedDim(X, reduction), "genesCoordinates"))[features, dims]
        featureData <- as.data.table(featureData, keep.rownames = "features")
        DIM1 <- colnames(featureData)[2]
        DIM2 <- colnames(featureData)[3]
        MCPlot <-
            plotReducedDim(X, ncomponents = dims, dimred = reduction, ...) +
            geom_point(data = featureData, mapping = aes_string(x = DIM1, y = DIM2, text = "features"), size = size.feature, shape = 4)
        if (as.text) {
            requireNamespace("ggrepel", quietly = TRUE)
            if (length(features) > 50) {
                if (menu(
                    choices = c("Yes", "No"),
                    title = "you are plotting as Text more than 50 genes, this can take a long time to plot, continue?"
                ) == 2) {
                    stop()
                }
            }
            MCPlot <-
                MCPlot + ggrepel::geom_text_repel(
                    data = featureData,
                    mapping = aes_string(
                        x = DIM1,
                        y = DIM2,
                        label = "features"
                    ),
                    size = size.feature.text,
                    point.padding = 0.2
                )
        }
        return(MCPlot)
    }