[BioC] Affymetrix Tag3

[Rafael A. Irizarry]

i dont think there is a package. below is a function i wrote for some 
folks here at jhu. you need the makecdfenv and affy packages for it to 
run. you need the TAG3 CDF and a TAG3info txt file with whats up and whats 
dn.  
if you, or anybody out there, 
wants to turn this into 
a formal package, let me know.


##THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND.
ReadTAG3 <- function(filename="Tag3_ID.txt", CDFfile = "TAG_3.CDF",
                     CELfiles = NULL,
                     compress.cel = FALSE,
                     compress.cdf = FALSE, verbose = T, chip.names = NULL,
                     rm.mask = FALSE, rm.outliers = FALSE, rm.extra = FALSE,
                     cdf.name = NULL,  ...)
{
  ##FIRST CHOOSE FILES
  ##with widget
  ##or without
  files <- list.files(...)
  if (is.null(CDFfile)) 
    CDFfile <- files[grep(".[cC][dD][fF]", files)]
  if (length(CDFfile) != 1) 
    stop(paste("CDFfile is not specified,exactly one CDF file must exist in path.\n"))
  if (is.null(CELfiles)) 
    CELfiles <- files[grep(".[cC][eE][lL]", files)]
  
  nchips <- length(CELfiles)
  if (nchips < 1) 
    stop(paste("CELfiles is not specified, at least one CEL file must exist in path.\n"))
  if (is.null(chip.names)) {
  chip.names <- CELfiles
  chip.names <- as.character(sapply(chip.names, function(x) strsplit(x,"\\.")[[1]][1]))
}
  else {
    if (length(chip.names) != nchips) {
      warning("Not the same number of chips than chip names. Assigning names from file.\n")
      chip.names <- CELfiles
      chip.names <- as.character(sapply(chip.names, function(x) strsplit(x,"\\.")[[1]][1])) 
    }
  }
  
    
  if (verbose) 
    cat("reading CDF file\n")
  cdf <- read.cdffile(CDFfile, compress = compress.cdf)
  if (verbose) 
    cat("processing information\n")
  
  info <- read.table(filename,header=T,sep="\t",as.is=T)
  upordown <- info[,"Plate"]
  upordown <- sapply(upordown,function(x)
                     if(length(grep("U",x)>0)) return("U")
                     else return("D"))

  geneNames <- info[,"Name"]
  tagNames <- info[,"Tag.name"]
  names(geneNames) <- tagNames
  names(upordown) <- tagNames
  
  probeNames <- cdf at name.levels[as.vector(cdf at name)]
  nrow <- dim(cdf at name)[1]
  ncol <- dim(cdf at name)[2]
  xs <- rep(0:(nrow-1), ncol)
  ys <- rep(0:(ncol-1), rep(nrow, ncol))
  Index <- probeNames%in%tagNames
  mmindex <- which(ys%%4==2 & Index)
  pmindex <- which(ys%%4==1 & Index)
  cpmindex <- pmindex+2*nrow
  cmmindex <- mmindex+2*nrow
  
  probeNames <- probeNames[pmindex]
  nprob <- length(pmindex)
  x <- xs[pmindex]
  y <- ys[pmindex]
  
  pm <- matrix(0, nprob, nchips)
  mm <- matrix(0, nprob, nchips)
  cpm <- matrix(0,nprob, nchips)
  cmm <- matrix(0,nprob, nchips)
  
  if (verbose) 
    cat("reading", nchips, "CEL files")
  for (i in 1:nchips) {
    aux <- as.vector(read.affybatch(filenames=CELfiles[i])@exprs)
    if (nrow * ncol != length(aux)) 
          stop(paste(CELfiles[i], "doesn't match with CDFfile\n"))
        pm[, i] <- aux[pmindex]
        mm[, i] <- aux[mmindex]
        cpm[, i] <- aux[cpmindex]
        cmm[, i] <- aux[cmmindex]
        if (verbose) 
          cat(".")
      }
  if (verbose) 
    cat("\npreparing probe level object\n")
  ## generate probenames for each probe using cdf at name.levels and probe.ids
  probe.ids <- as.vector(cdf at name)[pmindex]
  probe.names <- cdf at name.levels[probe.ids]


  colnames(pm) <- chip.names
  rownames(pm) <- probeNames
  colnames(mm) <-  chip.names
  rownames(mm) <-  probeNames
  colnames(cpm) <-  chip.names
  rownames(cpm) <- probeNames
  colnames(cmm) <-  chip.names
  rownames(cmm) <- probeNames
  
  
list(pm=pm,mm=mm,cpm=cpm,cmm=cmm,geneNames=geneNames[probeNames],
      probeNames=probeNames,upordown=upordown[probeNames],x=x,y=y)   
  
}









 On Mon, 19 Apr 2004, Julia Reid wrote:

> Can anyone point me to code designed specifically for analysis of 
> Affymetrix Tag3 array data?
> Much appreciated,
> Julia Reid
> 
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://www.stat.math.ethz.ch/mailman/listinfo/bioconductor
>