R/probBinom.R

In SIMD: Statistical Inferences with MeDIP-seq Data (SIMD) to infer the methylation level for each CpG site

#' Compute P-values for Medip-seq and MRE-seq data.
#'
#' @description Compute P-values.
#' @param t The real value for random variable according to dataset.
#' @param size1 The sum of Medip-seq real reads of the each CpG site for
#' control and treatment sample.
#' @param size2 The sum of MRE-seq real reads of the each CpG site for control
#' and treatment sample.
#' @param c1 The scaling factor for MeDip-seq data.
#' @param c2 The scaling factor for MRE-seq data.
#' @return p The P-values for testing the methylation expression levels for each
#' CpG sites.
#' @examples 
#' set.seed(1234)
#' t <- 0.1
#' size1 <- sample(1:1000, 1, replace=TRUE)
#' size2 <- sample(1:1000, 1, replace=TRUE)
#' c1 <- 1
#' c2 <- 2
#' result <- probBinom(t, size1, size2, c1, c2)
#' @export


probBinom <- function(t, size1, size2, c1, c2){
    prob1 <- 1/(1 + c1)
    prob2 <- 1/(1 + c2)
    i <- 0:size1
    j <- size1:0
    pij <- dbinom(i, prob=prob1, size=size1)
    r <- 0:size2
    k <- size2:0
    prk <- dbinom(r, prob=prob2, size=size2)
    ik <- outer(i, k)
    jr <- outer(j, r)
    ijrk <- c1*ik-c2*jr
    prob <- outer(pij, prk)
    ir <- which(abs(ijrk)>=t, arr.ind=TRUE)
    p <- sum(prob[ir])
    return(p)
}