AMR/R/mo_matching_score.R

64 lines
3.9 KiB
R
Executable File

# ==================================================================== #
# TITLE #
# Antimicrobial Resistance (AMR) Analysis #
# #
# SOURCE #
# https://github.com/msberends/AMR #
# #
# LICENCE #
# (c) 2018-2020 Berends MS, Luz CF et al. #
# #
# This R package is free software; you can freely use and distribute #
# it for both personal and commercial purposes under the terms of the #
# GNU General Public License version 2.0 (GNU GPL-2), as published by #
# the Free Software Foundation. #
# #
# We created this package for both routine data analysis and academic #
# research and it was publicly released in the hope that it will be #
# useful, but it comes WITHOUT ANY WARRANTY OR LIABILITY. #
# Visit our website for more info: https://msberends.github.io/AMR. #
# ==================================================================== #
#' Calculate the matching score for microorganisms
#'
#' This helper function is used by [as.mo()] to determine the most probable match of taxonomic records, based on user input.
#' @param x Any user input value(s)
#' @param fullname A full taxonomic name, that exists in [`microorganisms$fullname`][microorganisms]
#' @param uncertainty The level of uncertainty set in [as.mo()], see `allow_uncertain` in that function (here, it defaults to 1, but is automatically determined in [as.mo()] based on the number of transformations needed to get to a result)
#' @details The matching score is based on four parameters:
#'
#' 1. A human pathogenic prevalence \eqn{P}, that is categorised into group 1, 2 and 3 (see [as.mo()]);
#' 2. A kingdom index \eqn{K} is set as follows: Bacteria = 1, Fungi = 2, Protozoa = 3, Archaea = 4, and all others = 5;
#' 3. The level of uncertainty \eqn{U} that is needed to get to a result (1 to 3, see [as.mo()]);
#' 4. The [Levenshtein distance](https://en.wikipedia.org/wiki/Levenshtein_distance) \eqn{L} is the distance between the user input and all taxonomic full names, with the text length of the user input being the maximum distance. A modified version of the Levenshtein distance \eqn{L'} based on the text length of the full name \eqn{F} is calculated as:
#'
#' \deqn{L' = F - \frac{0.5 \times L}{F}}{L' = F - (0.5 * L) / F}
#'
#' The final matching score \eqn{M} is calculated as:
#' \deqn{M = L' \times \frac{1}{P \times K} * \frac{1}{U}}{M = L' * (1 / (P * K)) * (1 / U)}
#'
#' @export
#' @examples
#' as.mo("E. coli")
#' mo_uncertainties()
mo_matching_score <- function(x, fullname, uncertainty = 1) {
# fullname is always a taxonomically valid full name
levenshtein <- double(length = length(x))
if (length(fullname) == 1) {
fullname <- rep(fullname, length(x))
}
if (length(x) == 1) {
x <- rep(x, length(fullname))
}
for (i in seq_len(length(x))) {
# determine Levenshtein distance, but maximise to nchar of fullname
levenshtein[i] <- min(as.double(utils::adist(x[i], fullname[i], ignore.case = FALSE)),
nchar(fullname[i]))
}
# self-made score between 0 and 1 (for % certainty, so 0 means huge distance, 1 means no distance)
dist <- (nchar(fullname) - 0.5 * levenshtein) / nchar(fullname)
prevalence_kingdom_index <- tryCatch(MO_lookup[match(fullname, MO_lookup$fullname), "prevalence_kingdom_index", drop = TRUE],
error = function(e) rep(1, length(fullname)))
dist * (1 / prevalence_kingdom_index) * (1 / uncertainty)
}