R/mylogLikePoisMixObs.R

In HTSCluster: Clustering High-Throughput Transcriptome Sequencing (HTS) Data

mylogLikePoisMixObs <- function(y, conds, s, lambda, pi) 
{
  g <- length(pi)
  n <- dim(y)[1]
  cols <- dim(y)[2]
  nas <- 0
  y <- matrix(y, nrow = n, ncol = cols)
  mean <- PoisMixMean(y=y, g=g, 
                           conds=conds, s=s, lambda=lambda)
  logLike <- rep(0, n)
  index <- 1:g
  epsilon <- exp(-720)
  thresh <- -745
  nn <- 0
  logpi <- log(pi)
  ef <- matrix(logpi, nrow = n, ncol = g, byrow = T)
  for (k in 1:g) {
    ef[, k] <- ef[, k] + rowSums(dpois(y, mean[[k]], log = T))
  }
  efmax <- apply(ef, 1, max)
  ef <- ef - efmax
  logLike <- efmax + log(rowSums(exp(ef)))
  return(list(ll = logLike))
}