R/loadDataset.R

In ilyessr/conclus: ScRNA-seq Workflow CONCLUS - From CONsensus CLUSters To A Meaningful CONCLUSion

.checkLoadedDataParams <- function(file, columnID, header, sep, dec, type){

    if(!is.data.frame(file) && !file.exists(file))
        stop("'file' parameter should be a path of existing file.")

    if (!is.logical(header))
        stop("'header' parameter should be a boolean. Set TRUE if the first ",
                "row of the table corresponds to the column names, and FALSE ",
                "if it doesn't.")

    if (!is.character(sep) || !sep %in% c(" ", ",", "\t", ";"))
        stop("'sep' parameter should be the character used in the table ",
                "to separate the fields. Usually it's ' ' ';' ',' or '\t'.")

    if (!is.character(dec) || !dec %in% c(".", ","))
        stop("'dec' parameter should be the character used in the table ",
                "for decimal points, usually '.' or ',' .")

    if (!type %in% c("coldata", "countMatrix", "rowdata"))
        stop("'type' parameter should be the string 'coldata' 'rowdata' ",
                "or 'countMatrix'. ")

    if(!is.character(columnID) && isTRUE(type == "coldata"))
        stop("'columnCells' parameter should be the name of the column",
            " containing cell names/id")

    if(!is.character(columnID) && isTRUE(type == "rowdata"))
        stop("'columnGenes' parameter should be the name of the column",
                " containing gene SYMBOLS. ")
}


.checkCountMatrix <- function(df){
    if(ncol(df) < 100)
        stop("Not enough cells in the count matrix. There ",
            "should be at leat 100 cells. ", "The current count matrix ",
            "contains ", ncol(df), " cells.\n")
    
}


#' loadDataOrMatrix
#'
#' @description
#' This function allows to import the coldata, rowData or the countMatrix. It
#' formats each type of data to follow the requirements of CONCLUS.
#'
#' @usage
#' loadDataOrMatrix(file, type, columnID=NULL, header=TRUE, sep='\t', dec=".",
#'                     ignoreCellNumber=FALSE)
#'
#' @param file Path to the rowData, colData or Matrix.
#' @param type Values should be "coldata", "rowdata", or "countMatrix".
#' @param columnID For row and col data, column name containing cells/genes
#' names/id. Should be the same names of genes or cells that in the count
#' matrix. Should not be used when importing a matrix. Default=NULL.
#' @param header Set TRUE if the first row of the table corresponds to the
#' column names, and FALSE if it doesn't. Default=TRUE.
#' @param sep Character used in the table to separate the fields.
#' Usually it's ' ' ';' ',' or '\\t'. Default='\\t'.
#' @param dec Character used in the table for decimal points.
#' Usually '.' or ',' . Default=".".
#' @param ignoreCellNumber  CONCLUS needs a large number of cells to collect
#' statistics, so we recommend using CONCLUS if you have at least 100 cells.
#' If you still want to use a count matrix with a number of cells inferior 
#' to 100 cells, set ignoreCellNumber=TRUE. Default=FALSE.
#' 
#' @return The formatted row, col data or the matrix.
#' @export loadDataOrMatrix
#'
#' @examples
#'
#' ## ColData
#' coldataPath <- system.file("extdata/colData.tsv", package="conclus")
#' loadDataOrMatrix(file=coldataPath, type="coldata", columnID="cell_ID")
#'
#' ## RowData
#' rowdataPath <- system.file("extdata/rowData.tsv",  package="conclus")
#' loadDataOrMatrix(file=rowdataPath, type="rowdata", columnID="gene_ID")
#'
#' ## Count matrix
#' countmatrixPath <- system.file("extdata/countMatrix.tsv", package="conclus")
#' countMatrix <- loadDataOrMatrix(file=countmatrixPath, type="countMatrix",
#'                                 ignoreCellNumber=TRUE)
#'
#'
#' @author Ilyess RACHEDI and Nicolas DESCOSTES
#' @importFrom utils read.delim
#'
loadDataOrMatrix <- function(file, type, columnID=NULL, header=TRUE, sep='\t',
        dec=".", ignoreCellNumber=FALSE){

    .checkLoadedDataParams(file, columnID, header, sep, dec, type)

    if(!is.data.frame(file))
        if(isTRUE(all.equal(type, "countMatrix")))
            df <- read.delim(file=file, header=header, sep=sep, dec=dec,
                    na.strings=c("", "NA",  "<NA>"), stringsAsFactors=FALSE,
                    row.names=1)
        else
            df <- read.delim(file=file, header=header, sep=sep, dec=dec,
                    na.strings=c("", "NA",  "<NA>"), stringsAsFactors=FALSE)
    else
        df <- file
    
    if (isFALSE(ignoreCellNumber) &&  isTRUE(all.equal(type, "countMatrix"))) 
        .checkCountMatrix(df)
    
    if(isTRUE(all.equal(type, "coldata")))
        refColName <- "cellName"
    else if(isTRUE(all.equal(type, "rowdata")))
        refColName <- "nameInCountMatrix"
    else
        return(as.matrix(df))

    if(isFALSE(columnID %in% colnames(df)))
        stop("There is no column '", columnID, "' in the submitted ", "data")
    
    if(isTRUE(any(is.na(df[, columnID]))))
        stop("There are some NA values in the column you choose. ",
            "Please fill theses values or choose another column.")
    
    if(isFALSE(refColName %in% colnames(df))){
        ## Change the column name
        names(df)[names(df) == columnID] <- refColName
        ## Re-order the columns
        df <- df[, c(refColName, colnames(df)[!colnames(df) %in% refColName])]
        }

    if(length(unique(df[, refColName])) != nrow(df))
        stop("IDs should be unique. Please check the selected column.")
    else
        rownames(df) <- df[[refColName]]

    return(df)
}