R/get_symbol_exprSet.R

In jmzeng1314/humanid: gene id annotation

#' extract the expression matrix from a eSet object
#'
#' not only extract the expression matrix, but change the probeset ID to gene HUGO symbol ,and remove the duplicated genes .
#'
#' @param eSet  A expresstionSet,S4 object in R, which contain: assayData,phenoData,featureData
#' @param platformDB which microarray platform does the eSet use,such as  illuminaHumanv4.db, hgu133plus2.db
#' @param filter whether to filter genes show very low expression values, default:TRUE
#' @return expression matrix, rowname is the gene HUGO symbol
#' @export
#' @keywords get_symbol_exprSet
#' @examples
#' #' exprSet <- get_symbol_exprSet(K27M_WT_eSet)
#'
#'
get_symbol_exprSet <- function(eSet, platformDB = "hgu133plus2.db", filter = TRUE) {
    ## you must make sure that the eSet is a standard format, eg: read the cel files by Affy package
    library(platformDB, character.only = TRUE)
    library(annotate)
    probeset <- featureNames(eSet)
    exprSet <- exprs(eSet)
    exprSet <- na.omit(exprSet)
    
    probe2symbol_df <- toTable(get(paste0(sub(".db", "", platformDB), "SYMBOL")))
    
    exprSet = as.data.frame(exprSet)
    exprSet$probe_id = rownames(exprSet)
    
    tmp <- merge(probe2symbol_df, exprSet, by = "probe_id")
    dim(tmp)
    tmp <- tmp[, -1]
    exprSet_rmdup = rmDupID(tmp)
    dim(exprSet_rmdup)
    if (filter) {
        keepProbe <- apply(exprSet_rmdup, 1, function(x) all(x > 1))
        exprSet_rmdup <- exprSet_rmdup[as.logical(keepProbe), ]
        dim(exprSet_rmdup)
        
    }
    
    if (mean(rowMeans(exprSet_rmdup, na.rm = T), na.rm = T) > 20) 
        exprSet_rmdup = log2(exprSet_rmdup)  ## based on 2
    
    # boxplot(exprSet_rmdup,las=2) exprSet_rmdup <- na.omit(exprSet_rmdup)
    return(exprSet_rmdup)
    
}