agilentTxtWorkflow: Implement workflow on Agilent Gene Expression or Array CGH...
In Sage-Bionetworks/mGenomics: mGenomics

Description Usage Arguments Details Value Author(s) Examples

This function allows the user to implement supervised or unsupervised workflows for Agilent data.

1 2	agilentTxtWorkflow(rawDataDir, bio.var=NULL, adj.var=NULL, int.var=NULL, rm.adj=TRUE, verbose=TRUE, logFile=TRUE)

`rawDataDir`	Character string describing the directory in which we can find tab-delimited text files containing raw data from an Agilent microarray experiment. Other files (including other tab-delimiated text files) are allowed in the directory and recursive search is used to search any directories within directories.
`bio.var`	Model matrix parameterizing the biological variables of interest. Note this option can only be set if it applies to all platforms that will be processed by the workflow. If multiple platforms exist we recommend first splitting the data into directories based on their platform and then running the workflow on each individually.
`adj.var`	Model matrix parameterizing the adjustment variables. Note this option can only be set if it applies to all platforms that will be processed. See note above on how to proceed if data from multiple platforms exists.
`int.var`	A data frame with number of rows equal to the number of samples, with the unique levels of intensity-dependent effects. Each column parameterizes a unique source of intensity-dependent effect (e.g., array effects for column 1 and dye effects for column 2). Note this option can only be set if it applies to all platforms that will be processed. See note above on how to proceed if data from multiple platforms exists.
`rm.adj`	A logical scalar. If set to 'TRUE' the influence of the adjustment variables will be removed from the normalized data.
`verbose`	A logical scalar. If set to 'TRUE' various messages will be passed to the R console providing updates for each workflow.
`logFile`	Used to create (if set equal to TRUE) or define a logFile (if set to name of an existing file).

This function implements supervised and unsupervised workflows for Agilent data. The normalization is based on the supervised normalization of microarrays (snm) framework. The unsupervised workflows use a model containing a probe intercept term and intensity-dependent effects for dye and array to normalize the data. Supervised workflows can be built by specifying the biological, adjustment, or intensity-dependent variables, but only if the data in the specified directory comes from a single platform.

Note that this workflow defines the data for the Cy3 (green) or Cy5 (red) channels as the gProcessedSignal and rProcessedSignal values.

More information is available on the wiki for the metaGenomics project: http://sagebionetworks.jira.com/metaGenomics/WIKI/home

A list of lists, one for each platform processed by the workflow. The objects of the list that is returned are named according to the platform type. A list for a given platform contains the normalized data as a matrix in an eset as well as an empty list. In the future this list will be poulated with statistics useful for donwstream analysis of the data. Note that for data obtained from experiments that use both green and red channels we return the data matrices so that the green channels are in columns 1 to n, where n is the # of distinct microarrays measuring that platform, and the red channels are in columns n+1 to 2n. In general if the green channel is in column j for the jth sample, then its corresponding red channel is in column n+j.

Note that we have encountered many strange features of Agilent .TXT files that make them difficult to process. Given this, we developed this software to be robust to improperly formatted input data. When we encounter such a file its corresponding columns in the data matrices are set equal to NA for every probe in every channel. These columns are returned to the user, but do not influence the normalization procedure.

Brig Mecham <brig.mecham@sagebase.org>

##---- Should be DIRECTLY executable !! ----
##-- ==>  Define data, use random,
##--	or do  help(data=index)  for the standard data sets.

## The function is currently defined as
function (rawDataDir, bio.var, adj.var, int.var, rm.adj, verbose) 
{
    all.txt.files <- list.files(path = rawDataDir)[grep(".txt", 
        list.files())]
    if (length(all.txt.files) == 0) {
        cat("No Agilent Platforms Detected\n")
        fits <- "No Agilent Platforms Detected\n"
        class(fits) <- "try-error"
        return(fits)
    }
    platforms <- sapply(all.txt.files, getAgilentPlatform)
    if (platforms == "No Agilent Platforms Detected\n") {
        fits <- platforms
    }
    else {
        agilent.files <- as.numeric(which(!is.na(platforms)))
        all.agilent.files <- all.txt.files[agilent.files]
        platforms <- platforms[agilent.files]
        platforms <- standardizePlatforms(platforms)
        platform2file <- split(all.agilent.files, platforms)
        fits <- list()
        for (i in 1:length(platform2file)) {
            cat("Platform:", names(platform2file)[i], "\n")
            files <- platform2file[[i]]
            cat("Number of Samples:", length(files), "\n")
            obj <- loadAgilentFiles(files)
            if (length(files) > 1) {
                cat("Normalizing Data\n")
                platformObj <- normalizeAgilent(obj)
            }
            else {
                if (sum(!is.na(obj$red)) > 0) {
                  is.red = TRUE
                }
                if (sum(!is.na(obj$green)) > 0) {
                  is.green = TRUE
                }
                if (is.red & is.green) {
                  data <- cbind(obj$green, obj$red)
                }
                if (is.red & !is.green) {
                  data <- cbind(obj$red)
                }
                if (!is.red & is.green) {
                  data <- cbind(obj$green)
                }
                platformObj <- list(data = data, probePValues = obj$probePValues)
            }
            if (length(platform2file) > 1) {
                cat("---------------------------------------------------\n")
            }
            fits[[i]] <- platformObj
            nms <- names(platform2file)[i]
            nms <- gsub("-", ".", nms)
            names(fits)[i] <- nms
        }
    }
    fits
  }