R/dimred_tsne.R

In compbiomed/animalcules: Interactive microbiome analysis toolkit

#' Dimensionality reduction through t-SNE
#'
#' @param MAE A multi-assay experiment object
#' @param tax_level The taxon level used for organisms
#' @param color A condition to color data points by e.g. "AGE"
#' @param shape A condition to shape data points by e.g. "SEX"
#' @param k Plot dimensions e.g. c("2D","3D")
#' @param initial_dims The number of dimensions to use in reduction method
#' @param perplexity Optimal number of neighbors
#' @param datatype Datatype to use e.g. c("logcpm", "relabu", "counts")
#' @param tsne_cache Pass the cached data back into the function
#' @return A list with a plotly object and cached data
#'
#' @examples
#' data_dir <- system.file("extdata/MAE.rds", package = "animalcules")
#' toy_data <- readRDS(data_dir)
#' results <- dimred_tsne(toy_data,
#'   tax_level = "phylum",
#'   color = "AGE",
#'   shape = "GROUP",
#'   k = "3D",
#'   initial_dims = 30,
#'   perplexity = 10,
#'   datatype = "logcpm"
#' )
#' results$plot
#'
#' @import dplyr
#' @import plotly
#' @import tsne
#' @import magrittr
#' @import reshape2
#' @import MultiAssayExperiment
#'
#' @export
dimred_tsne <- function(MAE,
    tax_level,
    color,
    shape = NULL,
    k = c("2D", "3D"),
    initial_dims = 30,
    perplexity = 10,
    datatype = c("logcpm", "relabu", "counts"),
    tsne_cache = NULL) {
    # Extract data
    microbe <- MAE[["MicrobeGenetics"]]
    # host <- MultiAssayExperiment::experiments(MAE)[[2]]
    tax_table <- as.data.frame(rowData(microbe)) # organism x taxlev
    sam_table <- as.data.frame(colData(microbe)) # sample x condition
    counts_table <-
        as.data.frame(assays(microbe))[, rownames(sam_table)] #organism x sample
    if (is.null(tsne_cache)) {
        # Default variables
        k <- ifelse(match.arg(k) == "2D", 2, 3)
        datatype <- match.arg(datatype)
        df <- counts_table %>%
            # Sum counts by taxon level
            upsample_counts(tax_table, tax_level) %>%
            # Choose data type
            {
                if (datatype == "relabu") {
                    counts_to_relabu(.)
                } else if (datatype == "logcpm") {
                    counts_to_logcpm(.)
                } else {
                    .
                }
            } %>%
            # Fix constant/zero row
            {
                if (sum(base::rowSums(as.matrix(.)) == 0) > 0) {
                    . <- .[-which(base::rowSums(as.matrix(.)) == 0), ]
                } else {
                    .
                }
            } %>%
            # Transpose
            t()
        
        # t-SNE
        df.tsne <-
            tsne(scale(df), 
                k = k, 
                initial_dims = initial_dims, 
                perplexity = perplexity)
        rownames(df.tsne) <- rownames(df)
        if (k == 2) {
            colnames(df.tsne) <-
                c("X", "Y")
        } else {
            colnames(df.tsne) <- c("X", "Y", "Z")
        }
    } else {
        df.tsne <- tsne_cache
        k <- ncol(df.tsne)
    }
    
    # Merge in covariate information
    if (!is.null(shape)) {
        df.tsne.m <-
            merge(df.tsne, sam_table[, c(color, shape),
                drop = FALSE
            ], by = 0, all = TRUE)
        
        # When shape is required
        # Bypass duplicate colnames if color == shape
        shape <- colnames(df.tsne.m)[ncol(df.tsne.m)]
        df.tsne.m[[shape]] <- as.factor(df.tsne.m[[shape]])
    } else {
        df.tsne.m <-
            merge(df.tsne, sam_table[, color, drop = FALSE], by = 0, all = TRUE)
        shape <- "shape" # Referenced by plotly later
        df.tsne.m[[shape]] <- 1 # Constant results in omitting shape
    }
    
    # Plotly | Scatterplot
    if (k == 2) {
        # 2D Plot
        p <- plot_ly(df.tsne.m,
            x = as.formula("~X"),
            y = as.formula("~Y"),
            mode = "markers",
            color = as.formula(paste("~", color, sep = "")),
            symbol = as.formula(paste("~", shape, sep = "")),
            type = "scatter",
            text = df.tsne.m$Row.names,
            marker = list(size = 10)
        )
    } else {
        # 3D Plot
        p <- plot_ly(df.tsne.m,
            x = as.formula("~X"),
            y = as.formula("~Y"),
            z = as.formula("~Z"),
            mode = "markers",
            color = as.formula(paste("~", color, sep = "")),
            symbol = as.formula(paste("~", shape, sep = "")),
            symbols = c(
                "circle", "square", "diamond",
                "cross", "square-open", "circle-open",
                "diamond-open", "x"
            ),
            type = "scatter3d",
            text = df.tsne.m$Row.names,
            marker = list(size = 6)
        )
    }
    p$p <- NULL # To suppress a shiny warning
    return(list(plot = p, data = df.tsne))
}