R/makeDist.R

In Scavetta/ComPrAn: Complexome Profiling Analysis package

#' Make disstance matrix
#'
#' This function calculates distance matrix for a data frame, column by column
#' requires uncenteredCor function to work
#'
#' @param df data frame, contains columns:
#'              `Protein Group Accessions` character
#'              `Protein Descriptions` character
#'               isLabel character ('TRUE'/'FALSE')
#'               columns 1 to n, numeric, n is the total number of 
#'               fractions/slices, each of this columns 
#'               contains `Precursor Area` values in a given fraction(columns) 
#'               for a protein(rows)
#' @param centered centered: logical,if TRUE return dist matrix based on 
#'                 centered Pearson correlation (uses R cor() function, fast)
#'                 ,if FALSE return dist matrix based on uncentered Pearson
#'                 correlation (uses custom uncenteredCor() function, slow)
#'
#' @importFrom stats as.dist cor
#'
#' @return matrix
makeDist <- function(df,centered = FALSE){
    if (centered){
        answer <- (1 - cor(df))
    } else {
        nc = ncol(df)
        answer = matrix(ncol=nc,nrow=nc)
        for (i in seq_len(nc)){
            for (j in seq_len(nc)){
                answer[i,j] = 1-(uncenteredCor(df[,i],df[,j]))
            }
        }
        rownames(answer) <- colnames(df)
        colnames(answer) <- colnames(df)
    }
    return(as.dist(answer))
}