--- a
+++ b/R/readDACData.R
@@ -0,0 +1,37 @@
+#' Read DLBCL class files
+#'
+#' Functions to read the internal data files of classified CEL files using
+#' the DLBCL Automatic Classifier (DAC) [1]. These were manually supplied to the
+#' DAC Windows interface.
+#'
+#' @param geo_nbr A \code{character} string giving the GSE number.
+#' @return A \code{data.frame} of the predicted classes and the class
+#'   probabilities. Returns \code{NULL} if \code{geo_nbr} is not found.
+#' @author
+#'   Anders Ellern Bilgrau \cr
+#'   Steffen Falgreen
+#' @references
+#'   [1] Care, M. A., Barrans, S., Worrillow, L., Jack, A., Westhead, D. R., &
+#'       Tooze, R. M. (2013). A microarray platform-independent classification
+#'       tool for cell of origin class allows comparative analysis of gene
+#'       expression in diffuse large B-cell lymphoma.
+#'       PloS One, 8(2), e55895. doi:10.1371/journal.pone.0055895
+#' @note
+#' The GSE11318 dataset was not classified using the DAC.
+#' @examples
+#' head(DLBCL_overview[, -c(3,4,6)])
+#'
+#' head(readDACData("GSE19246"))
+#'
+#' readDACData("UNSUPPORTED_STUDY")
+#' @export
+readDACData <- function(geo_nbr) {
+  extdir <- system.file("extdata", package = "DLBCLdata")
+  files <- list.files(extdir, pattern = "^predictions-.+\\.txt$",
+                      full.names = TRUE)
+  file <- grep(geo_nbr, files, value = TRUE)
+  dat <- lapply(file, read.table, header = TRUE)
+  names(dat) <- gsub("predictions-|\\.txt$", "", basename(file))
+  ans <- do.call(rbind, dat)
+  return(ans)
+}