R/features.r

In perishky/meffil: Efficient algorithms for DNA methylation

#' Get a list of microarray features from a predefined feature set.
#'
#' @param featureset A name returned by \code{\link{meffil.list.featuresets()}} (Default: "450k").
#' @return A data frame listing all features in the feature set.
#' @export
meffil.get.features <- function(featureset="450k")
    meffil.featureset(featureset)

# Get a subset of features from a predefined feature set.
#
# @param featureset A name returned by \code{\link{meffil.list.featuresets()}}.
# @param expr An expression using column names in the feature set data frame
# to select specific features.
# @return A data frame listing all features satisfying \code{expr}.
#
# @examples
# ## Obtain a list of features measuring methylation using Type II probes.
# filter.features("450k", target=="methylation" & type == "ii")
# 
filter.features <- function(featureset, expr) {
    features <- meffil.get.features(featureset)
    sat <- eval(substitute(expr), features)
    idx <- which(sat)
    features[idx,]
}

# select a compatible featureset for the set of features.
guess.featureset <- function(features) {
    featuresets <- meffil.list.featuresets()
    featuresets <- sapply(featuresets, function(x) {
        meffil.get.features(x)$name
    }, simplify=F)
    pto <- sapply(featuresets, function(fset) mean(features %in% fset, na.rm=T))
    pfrom <- sapply(featuresets, function(fset) mean(fset %in% features, na.rm=T))
    names(featuresets)[which.max(pfrom*pto)]
}