R/operations_on_scores.R

In charles-bernard/mage: Mine Associated Gene Expressions (from single-cell Rna Seq data)

# mage R package

# This file regroups functions to:
#   1. Filter the matrix based on a threshold of significance
#   2. Standardize the matrix (using zscore)


#' @title filter_scores
#'
#' @description
#' Filters a table of scores based on a threshold of significance
#' defined either by a MIC score, or by a p-value
#'
#' @param x data.frame: the table of scores
#'
#' @param on variable used for the threshold
#' (accepts only \code{"MIC"} or \code{"pval"})
#'
#' @param pval numeric vector of p-values corresponding to
#' the MIC scores in \code{x} (only if \code{on = "pval"})
#'
#' @param thresh value of the threshold
#'
#' @return
#' Returns a filtered table of scores
#'
#' @examples
#' \dontrun{
#' scores_tab <- compute_scores(my_gene_exp_matrix, n.cores = 6)
#' pvalues <- assign_pval(scores_tab$`MIC (strength)`, nb_cells = 96)
#' signif_scores_tab <- filter_scores(scores_tab, on = "pval",
#'                                    pval = pvalues, threshold = 0.05)
#' }
#'
#' @details
#' if \code{on = "MIC"}, \code{filter_scores} will retain any association whose MIC >= \code{threshold}
#'
#' else if \code{on = "pval"}, \code{filter_scores} will retain any association whose p-value <= \code{threshold}
#'
filter_scores <- function(x,
                   on = 'MIC',
                   pval = NULL,
                   thresh = 0.4) {
  # To do: check all conditions

  if(on == "MIC") {
    if(thresh < 0 || thresh > 1) {
      stop("A MIC score is defined in [0;1]");
    }
    retained_ix <- which(x[, 3] >= thresh);
  } else if(on == "pval") {
    if(is.null(pval)) {
      stop("argument pval missing");
    }
    if(length(pval) != nrow(x)) {
      stop("length(pval) must be equal to nrow(x)");
    }
    retained_ix <- which(pval <= thresh);
  } else {
    stop("on argument must be set to either \"MIC\" or \"pval\"");
  }

  return(x[retained_ix, ]);
}

#' @title standardize_scores
#'
#' @description
#' Standardize each column of the table of scores via the Zscore: \deqn{x = ( x - mean(column) ) / ( std(column) )}
#'
#' @param x data.frame, table of scores
#'
#' @return
#' returns a standardized table of scores
standardize_scores <- function(x) {
  std_x <- scale(x[, 3:ncol(x)]);

  possible_NA_col <- apply(std_x, 2, function(i) any(is.na(i)));
  if(length(which(possible_NA_col == TRUE)) > 0) {
    std_x <- std_x[, -(which(possible_NA_col == TRUE))];
  }

  std_tab <- data.table(x[, 1:2], std_x);

  return(std_tab);
}