R/treebuilding.r

In zachcp/phyloseq-tools: convenience functions for working with phyloseq

########################################################################################
#
# Note : This namespace has orphaned funtions not ready for primetime or really anytime.
#
########################################################################################

#'
#' Get the  unique values from a column and return a simple phylogenetic tree 
#' 
#' This function is intened for use with a dataframe of cluster information where
#' each row is a sequence/OTU and each column is a level of similariy (percent identity)
#' this function will return a tree based on the lowest percent identity column and
#' we will build off of it.
#' @importFrom stringr str_c
#' @importFrom ape read.tree
column_to_tree <- function(df, colname){
  #get unique names, make newick string and tree
  concatnames <- unique(df[[colname]]) %>% str_c(collapse = ",")
  tree = paste("(", concatnames,");")
  read.tree(text = tree)
}

#'
#' add a tree to a base tree at tipname
#' 
#' @importFrom ape bind.tree
#' @return phylo object
#' @keywords internal
add_tree_at_tip <-function(basetree, incomingtree, tipname){
  bind.tree(basetree, incomingtree, where=which(basetree$tip.label==tipname))
}
#'
#' get values from adjacent row
#' 
#' in a table representing phylogenetic relationships
#' each column of the table correspods to one more 
#' level of clustering. this will get the values of
#' the "higher" column, based on shared rows with the 
#' "lowercolumn"
#' 
#' @return phylo object
#' @importFrom ape read.tree
#' @keywords internal
get_next_order_tree <- function(df, lowidcol, highidcol, val){
  newvals <- df[df[[lowidcol]] == val,][[highidcol]]
  vals    <- str_c(unique(newvals),collapse=",")
  tree    <- paste("(", vals,");")
  return(read.tree(text=tree))
}
#'
#' make a tree from a dataframe
#' @export
make_tree <- function(df){
  numcols <- length(names(df))
  # use last column for base tree and loop backwards
  basetree <- column_to_tree(df, numcols) 
  for (i in c(numcols:1)){
    #loop through the columns and start with the second column
    if (i < numcols) {
      for (val in unique(df[[i+1]])){
        #get the values for the tree...
        incomingtree <- get_next_order_tree(df=df, 
                                            lowidcol= i+1,
                                            highidcol= i,
                                            val=val)
        #...and tack them onto the original tree.
        basetree <- add_tree_at_tip(basetree,incomingtree,val)
      }
    }
  }
  return(basetree)  
}