R/scaleNorm-methods.R

In katenambiar/pmpa: Peptide Microarray Pre-Processing and Analysis

#' Scale Normalisation
#' 
#' 
#' @param x MultiSet object with fMedian matrices 
#' in the assayData slot
#' @param offset numeric value added to raw signal intensity 
#' before background correction is implemented
#' @return MultiSet object with normalised 
#' foreground signal in the fMedian matrix
#'  
#' @exportMethod scaleNorm
#' @docType methods
#' @rdname scaleNorm-methods
setGeneric(
  name = "scaleNorm", 
  def = function(x, ...) standardGeneric("scaleNorm")
)

#' @rdname scaleNorm-methods
#' @aliases scaleNorm
setMethod(
  f = "scaleNorm",
  signature = "MultiSet",
  definition = function(x, controlID = NULL){
    
    if (is.null(controlID)){
      y <- apply(fg(x), 2, median, na.rm = TRUE)
      y <- y - exp(mean(log(abs(y))))
      z <- sweep(fg(x), 2, y)
      assayDataElement(x, "fMedian") <- z
      
      return(x)
      
    } else {
      y <- fg(x)[fData(x)$ID %in% controlID, ]
      y <- apply(y, 2, median, na.rm = TRUE)
      y <- y - exp(mean(log(abs(y))))
      z <- sweep(fg(x), 2, y)
      assayDataElement(x, "fMedian") <- z
      
      return(x)
    }
    
  }
)

#' @rdname scaleNorm-methods
#' @aliases scaleNorm
setMethod(
  f = "scaleNorm",
  signature = "ExpressionSet",
  definition = function(x, controlID = NULL){
    
    if (is.null(controlID)){
      y <- apply(exprs(x), 2, median, na.rm = TRUE)
      y <- y - exp(mean(log(abs(y))))
      z <- sweep(exprs(x), 2, y)
      exprs(x) <- z
      
      return(x)
      
    } else {
      y <- exprs(x)[featureNames(x) %in% controlID, ]
      y <- apply(y, 2, median, na.rm = TRUE)
      y <- y - exp(mean(log(abs(y))))
      z <- sweep(exprs(x), 2, y)
      exprs(x) <- z
      
      return(x)
    }
    
  }
)