R/pvalueWeightedEst.R

In Lululuella/OVESEG: OVESEG-test to detect tissue/cell-specific markers

#' p-value by weighted permutation scheme
#'
#' This function estimates p-values by aggregating weighted permutations.
#' @param tt a vector of test statistics.
#' @param ttperm a matrix of test statistics from permutations.
#'     Rows correspond to probes and columns to one permutation.
#' @param w a matrix containing weights for each spot in \code{ttperm}.
#'     Provided by \code{\link{postProbNull}}.
#' @details P-values are estimated by weightedly accumulating test statistics
#' from permutations that are larger than observations
#' @return  p-values.
#' @export
#' @examples
#' #generate some example data
#' t.obs <- rnorm(100)
#' t.perm <- matrix(rnorm(100*1000),nrow=100)
#' w <- matrix(runif(100*1000),nrow=100)
#'
#' pv <- pvalueWeightedEst(t.obs, t.perm, w)
pvalueWeightedEst <- function(tt, ttperm, w)
{
    tt <- c(tt)
    ttperm <- c(ttperm)
    w0 <- rep(1, length(tt))

    r0 <- rank(-c(tt, ttperm), ties.method = 'last')
    w1 <- c(w0, w)
    w1[r0] <- w1
    wrank <- cumsum(w1)

    r <- r0[seq_along(tt)]
    r2 <- rank(-tt, ties.method = 'last')
    r3 <- wrank[r] - r2
    pv <- r3/sum(w)
    return(pv)
}