R/correlations_to_densities.R

In scFeatureFilter: A correlation-based method for quality filtering of single-cell RNAseq data

.applyDensity <- function(x, n = 64, absolute_cc = TRUE) {
    if (absolute_cc) {
        myDens <- density(abs(x), n = n, from = 0, to = 1, cut = 0)
    } else {
        myDens <- density(x, n = n, from = -1, to = 1, cut = 0)
    }
    return(list(
        cor_coef = myDens$x,
        density = myDens$y
    ))
}

# suppress CHECK annoying handling of NSE
utils::globalVariables(c("window", "."))
#' Transform the correlation table to density distributions of correlation values
#'
#' Takes the output of \code{\link{correlate_windows}} and computes density
#' curves of correlation coefficient for each window comparison.
#'
#' @param df A data frame, usually the output of \code{\link{correlate_windows}}.
#'
#' @param n Resolution of the correlation density curve. Default to 64.
#'
#' @param absolute_cc Should the function use the absolute value of correlation coefficients?
#' Default to \code{TRUE} to simplify plots and avoid annoying, non-symmetrical,
#' near 0, shifts of distributions.
#'
#' @return A \code{\link{tibble}} with columns \code{bin}, \code{window}, \code{cor_coef} and \code{density}.
#'
#' @examples
#' library(magrittr)
#' expMat <- matrix(
#'     c(1, 1, 5,
#'       1, 2, 3,
#'       0, 1, 4,
#'       0, 0, 2),
#'     ncol = 3, byrow = TRUE, dimnames = list(paste("gene", 1:4), paste("cell", 1:3))
#' )
#'
#' calculate_cvs(expMat) %>%
#'     define_top_genes(window_size = 2) %>%
#'     bin_scdata(window_number = 1) %>%
#'     correlate_windows %>%
#'     correlations_to_densities
#'
#' @export
correlations_to_densities <- function(df, n = 64, absolute_cc = TRUE) {
    dplyr::group_by(df, bin, window) %>%
        dplyr::do(do.call(tibble::tibble, .applyDensity(.$cor_coef, n = n, absolute_cc = absolute_cc)))
}