R/get_relative_ont_level.R

In ovrhuman/HPOExplorer: Analysis and Visualisation of the Human Phenotype Ontology

#' Get relative ont level (for terms in a connected component of HPO subset)
#'
#' When plotting subsets of the HPO data as a network plot, there is often more than
#' one connected component to be plotted (i.e. sections of the graph are not connected
#' by any edges). To map the ontology level to the size of nodes (such that high up terms
#' are bigger than low level terms). This was made to find the relative ontology level
#' with respect to other terms in its connected component so that each connected section
#' of the plot would have the same size for its root node. It would also be possible
#' to just use absolute ontology level, so that "Phenotypic abnormality" would always
#' be the largest datapoint etc.
#'
#' @param phenotype HPO term Id <string>
#' @param phenoAdj A adjacency matrix (produced by the adjacency_matrix function)
#' @param hpo The HPO ontology data object
#'
#' @returns A integer representing the relative ontology level of a term within
#' a connected component of a subset of the HPO.
#' @export

get_relative_ont_level <- function (phenotype,phenoAdj,hpo){
  pos_parents = hpo$parents[phenotype]
  phenotypes = rownames(phenoAdj)
  paths = list()
  for (p in phenotypes){
    if (phenoAdj[p,phenotype] == 1) {
      if (p %in% pos_parents) {
        paths[p] = 1 + get_relative_ont_level(p,phenoAdj,hpo) # <- recursion
      }
    }
  }
  if (length(paths) == 0) {
    return (0)
  } else {
    parents = 0
    for (i in seq(length(paths))) {
      if (paths[[i]] > parents) {
        parents = paths[[i]]
      }
    }
  }
  return (parents)
}