R/utils.R

In CrossICC: An Interactive Consensus Clustering Framework for Multi-platform Data Analysis

# x is a list object only containing one element (used in apply)
check.eSet <- function(x) {
    if (is.data.frame(x)) {
        as.matrix(x)
    }
    if (!is.numeric(x[1])) {
        stop("The first element from your data is not numeric:\n
         Currently, we only support matrix (data.frame) with sample names as column names\n
         and feature names as row names.")
    } else {
        if (!is.matrix(x)) {
            as.matrix(x)
        } else {
            x
        }
    }
}

matrix.please <- function(x) {
    # Thanks to https://stackoverflow.com/a/25695207
    m <- as.matrix(x[, -1])
    rownames(m) <- x[, 1]
    m
}

get.max.var.rows <- function(x) {
    DT <- data.table(row.names(x), x, var = apply(x, 1, var))
    # Thanks to https://stackoverflow.com/a/29497254
    DT <- setDT(DT)[, .SD[which.max(var)], by = V1]
    m <- matrix.please(DT)
    m[, -dim(m)[2]]
}

# Thanks to https://stackoverflow.com/a/4752580 for this function
zero_range <- function(x, tol = .Machine$double.eps^0.5) {
    if (length(x) == 1) 
        return(TRUE)
    x <- range(x)/mean(x)
    isTRUE(all.equal(x[1], x[2], tolerance = tol))
}

# Thanks to https://support.bioconductor.org/u/3556/ at https://support.bioconductor.org/p/41288/#41289 for this statement.
merge.duplicates <- function(x, method = "median") {
    switch(method, median = matrix.please(aggregate(x, by = list(row.names(x)), FUN = median)), mean = matrix.please(aggregate(x, by = list(row.names(x)), 
        FUN = mean)), max = matrix.please(aggregate(x, by = list(row.names(x)), FUN = max)), min = matrix.please(aggregate(x, by = list(row.names(x)), 
        FUN = min)), var = get.max.var.rows(x))
}

add.jitter <- function(x) {
    t(apply(x, 1, function(x) if (zero_range(x) == TRUE) {
        jitter(x)
    } else {
        x
    }))
}


remove.all.same <- function(m) {
    keep <- apply(m, 1, function(x) length(unique(x[!is.na(x)])) != 1)
    m[keep, ]
}