R/outer2df.R

In lwaldron/doppelgangR: Identify likely duplicate samples from genomic or meta-data

.outer2df <- function  #going from matrices to two columns
### Output rows contain all possible
### pairwise combinations of x and y
(x, y,
 ### If x is a vector and y is a vector, return all or subsets of
 ### outer(x, y) depending on the values of bidirectional and diag.  If
 ### x is a matrix and y is unspecified, return the elements of x in
 ### the same order as would be returned if x=rownames and y=colnames.
 bidirectional = TRUE,
 ### If TRUE, include separate rows for elements i, j and j, i
 diag = TRUE
 ### If TRUE, include i, i elements.
) {
  bizarre.and.unlikely.separator = " as3a2s5df5hjnm4qwe2rxcvb "
  if (is(x, "matrix")) {
    output.samples <- x
  } else if (is.vector(x) & is.vector(y)) {
    output.samples <-
      outer(x, y, paste, sep = bizarre.and.unlikely.separator)
  } else{
    stop("Require either x to be a matrix, or both x and y to be vectors.")
  }
  if (!bidirectional & !diag) {
    output.samples <- output.samples[upper.tri(output.samples)]
  } else if (!bidirectional & diag) {
    output.samples <- output.samples[!lower.tri(output.samples)]
  } else if (bidirectional & !diag) {
    output.samples <-
      output.samples[upper.tri(output.samples) |
                       lower.tri(output.samples)]
  } else if (bidirectional & diag) {
    output.samples <- as.vector(output.samples)
  }
  if (any(grepl(bizarre.and.unlikely.separator, output.samples, fixed =
                TRUE))) {
    output <-
      data.frame(do.call(
        rbind,
        strsplit(output.samples, split = bizarre.and.unlikely.separator)
      ),
      stringsAsFactors = FALSE)
  } else{
    output <- output.samples
  }
  return(output)
  ### A two-column dataframe if x and y are vectors, or a vector if x is
  ### a matrix and y is NULL.
}