R/calc_AUC.R

In alex-kalinka-cruk/fgcQC: QC metric calculations for CRISPR screen data produced at the FGC

.get_auc <- function(scores, TP){
  # Taking all pairs of TP and FP, what is probability TP have a higher score than FP?
  score_fp <- scores[!TP]
  score_tp <- scores[TP]
  o <- outer(score_tp, score_fp, "-")
  return(mean((o>0) + 0.5*(o==0)))
}

#' calc_AUC
#'
#' Calculates AUROC in a data frame containing a 'TP' column.
#'
#' @param data A data frame containing a 'TP' column.
#' @param score_col A character string naming the column to be used as a score column.
#' @param group A character string naming a column by which to group 'data'. If `NULL` no grouping, defaults to `NULL`.
#'
#' @return A data frame containing an `AUC` column.
#' @author Alex T. Kalinka, \email{alex.kalinka@@cancer.org.uk}
#' @importFrom dplyr mutate summarise group_by sym
#' @importFrom magrittr %<>%
#' @export
calc_AUC <- function(data, score_col, group = NULL){
  if(!"TP" %in% colnames(data))
    stop("expecting to find a 'TP' column in 'data'")
  if(!score_col %in% colnames(data))
    stop(paste("expecting to find a",score_col,"column in 'data'"))

  tryCatch({
    if(!is.null(group)){
      if(!group %in% colnames(data))
        stop(paste("expecting to find a",group,"column"))
      grp <- dplyr::sym(group)
    }
    sc <- dplyr::sym(score_col)
    if(!is.null(group)){
      data %<>%
        dplyr::group_by(!!grp)
    }
    data %<>%
        dplyr::summarise(AUC = .get_auc(!!sc, TP))
  },
  error = function(e) stop(paste("calc_AUC: unable to calculate AUROC:",e))
  )
  return(data)
}