R/get_aldex_exp.R

In thomazbastiaanssen/Tjazi: Microbiome Oriented Compositional Data Toolkit

#' Extract the CLR-transformed values from the ALDEX2
#' @description Microbiome data is compositional. When compositional data is examined using non-compositional methods, many problems arise.
#' Performing a centered log-ratio transformation is a reasonable way to address these problems reasonably well. This particular method takes advantage of the excellent ALDEX2 package.
#'
#' @export
get_aldex_exp = function(clr, useMC = TRUE, verbose = TRUE){
  is.multicore = FALSE
  if ("BiocParallel" %in% rownames(installed.packages()) &
      useMC) {
    message("multicore environment is OK -- using the BiocParallel package")
    is.multicore = TRUE
  }
  else {
    if (verbose == TRUE)
      message("operating in serial mode")
  }


  nr <- ALDEx2::numFeatures(clr)
  rn <- ALDEx2::getFeatureNames(clr)
  cn <- ALDEx2::getSampleIDs(clr)

  if (is.multicore == TRUE)
    clr.list <- BiocParallel::bplapply(ALDEx2::getMonteCarloInstances(clr), function(m) {
      t(apply(m, 1, median))
    })
  if (is.multicore == FALSE)
    clr.list <- lapply(ALDEx2::getMonteCarloInstances(clr), function(m) {
      t(apply(m, 1, median))
    })


  clr.exp = data.frame(matrix(unlist(clr.list), nrow = nr, byrow = F),stringsAsFactors=FALSE)
  row.names(clr.exp) = rn
  colnames(clr.exp)  = cn

  return(clr.exp)
}