R/sort_counts.R

Defines functions sort_counts

Documented in sort_counts

#' Sort columns of a count matrix by rows in sample data
#'
#' \code{DESeqDataSetFromMatrix} requires count columns sorted by the rows in sample data and only checks
#' if sample row names match count column names (and tibbles do not have row names).  This matches column
#' names in a count table to the first column in sample data and reorders columns to match.
#'
#' @param count_tbl count matrix, data.frame or tibble
#' @param sample_tbl sample tabl with a column containing the columns names in count matrix
#' @param id position of column in sample data that matches column names in the
#'   count table, default is the first column
#'
#' @return A count matrix with columns re-ordered to match the sample table
#'
#' @author Chris Stubben
#'
#' @examples
#' \dontrun{
#' count_tbl <-  sort_counts(counts, samples)
#' }
#' @export

sort_counts <- function(count_tbl, sample_tbl, id=1){
   cname <- deparse(substitute(count_tbl))
   orig_count_tbl <- count_tbl
   if( class(count_tbl)[1] != "matrix"){
        count_tbl <- as_matrix(count_tbl)
        if( !is.numeric(count_tbl)) stop("Count table is not numeric.
  Counts should be the first option and you used ", shQuote(cname))
   }
   # match first column in sample data by default
   if( ncol(count_tbl) !=  nrow(sample_tbl) ){
         stop("count_tbl should have one more column than sample_tbl rows")
   }
   n <- match(sample_tbl[[id]], colnames(count_tbl ) )
   if(any(is.na(n))) stop( "Column names in count_tbl do not match sample names in column ", id)
   if(all(diff(n)==1) ){
       # message("counts are already sorted by samples")
       c1 <- orig_count_tbl
   }else{
      message("Reordering columns in counts to match samples")
      if( dplyr::is.tbl( orig_count_tbl)){
         c1 <- orig_count_tbl[, c(1, n+1) ]
      }else{
         c1 <- orig_count_tbl[, n ]
      }
   }
   c1
}
HuntsmanCancerInstitute/hciR documentation built on Nov. 15, 2019, 1:37 p.m.