dataPreprocessing

datenVorverarbeitung <- function(dataDirectory, pattern = "*"){
  # save current working directory to get there later
  currentwd = getwd()
  # change working directory to the dataDirectory passed in the function head
  setwd(dataDirectory)
  # save the filenames in a vector can be limited by pattern passed in the function head
  list = dir(pattern = pattern)

  # iterate over the list of filenames
  for (j in 1:length(list)){
    data = read.delim(list[j])
    # subtract one value of the other (green red)
    ch1 = data$CH1I - data$CH1B
    ch2 = data$CH2I - data$CH2B

    # replace le values with the lowest possible values (1)
    ch1 = replace(ch1, ch1<=0,1)
    ch2 = replace(ch2, ch2<=0,1)

    M = log2(ch1) - log2(ch2)
    A = (log2(ch1) + log2(ch2)) / 2

    regressionsGerade = loess(M~A)

    # within array normalisation
    M2 = M - predict(regressionsGerade, A)
    # centering
    M3 = M2 - median(M2)
    # scaling
    M4 = M3/mad(M3)

    # comcatenate the vectors
    if(j == 1){
      MM = M
      MM2 = M2
      MM3 = M3
      MM4 = M4
      files = list[j]
    }
    else {
      MM = cbind(MM, M)
      MM2 = cbind(MM2, M2)
      MM3 = cbind(MM3, M3)
      MM4 = cbind(MM4, M4)
      files = cbind(files, list[j])
    }
  }

  # plotting the vectors
  par(mfrow=c(1,4))
  colnames(MM) = files
  boxplot(data.frame(MM), main="before \"Within Array Normalisation\"", las=1, ylim=c(-15,15))
  colnames(MM2) = files
  boxplot(data.frame(MM2), main="after \"Within Array normalisation\"", las=1, ylim=c(-15,15))
  colnames(MM3) = files
  boxplot(data.frame(MM3), main="after \"Centering\"", las=1, ylim=c(-15,15))
  colnames(MM4) = files
  boxplot(data.frame(MM4), main="after \"Scaling\"", las=1, ylim=c(-15,15))

  # set cuttent working directory back to the previous state
  setwd(currentwd)
}