R/rulesTF.R

In gaiac/TFARM: Transcription Factors Association Rules Miner

#' Extracts a subset of rules that contain a certain transcription factor
#' (or a combination of transcription factors) in their left-hand-side.
#'
#' From a set of relevant association rules, only the ones containing \code{TFi}
#' in their left-hand-side are subsetted, together with their quality measures
#' of support, confidence and lift. The function is then used for the evaluation
#' of the importance distribution of \code{TFi}.
#'
#' @param TFi a string with the name of the transcription factor (or combination
#' of transcription factors) wanted in the left-hand-side of the rules to find.
#' @param rules a data.frame with association rules and their quality measures
#' of support, confidence and lift.
#' @param verbose a logical parameter. If \code{verbose = TRUE}, a warning
#' message is reported to the user when the set of rules containing \code{TFi}
#' is empty.
#'
#' @return A data.frame with association rules containing \code{TFi} in their
#' left-hand-side, with their quality measures of support, confidence and lift.
#' @export
#'
#' @examples
#' # Load r_TEAD4 from the data_man collection of datasets:
#' data('data_man')
#'
#' r_FOSL2 <- rulesTF('FOSL2=1', r_TEAD4, verbose=FALSE)
#'
rulesTF <- function(TFi, rules, verbose) {
    if (length(TFi) == 1) {
        TF_i <- TFi
    }
    if (length(TFi) > 1) {
        TF_i <- TFi[[1]]
        for (i in 2:length(TFi)) TF_i <- paste(TF_i, TFi[[i]], sep = ",")
    }
    n_rules <- dim(rules)[1]
    rules_subs <- NULL
    for (i in 1:n_rules) {
        if (all(items(TF_i) %in% items(rules$lhs[i])))
            rules_subs <- cbind(rules_subs, paste(rules$lhs[i]))
    }
    if (length(rules_subs) == 0) {
        if (verbose == "TRUE") {
            warning(paste('None of the rules contains', TF_i, sep=' '))
            return(NA)
        } else if (verbose == "FALSE")
            (return(NA))
    } else {
        n_subs <- length(rules_subs)
        n_all <- dim(rules)[1]
        all_TF <- matrix(0, n_subs, 5)
        all_TF <- data.frame(all_TF)
        colnames(all_TF) <- c("lhs", "rhs", "support", "confidence", "lift")
        for (i in 1:n_subs) {
            d <- items(rules_subs[i])
            for (j in 1:n_all) {
                c <- items(rules$lhs[j])
                if (all(c %in% d)) {
                    all_TF[i, 3:5] <- rules[j, c(3, 4, 5)]
                    all_TF[i, 2] <- paste(rules[j, 2])
                    all_TF[i, 1] <- paste(rules[j, 1])
                }
            }
        }

    }
    return(all_TF)
}