AMR/data-raw/reproduction_of_rsi_translation.R

# ==================================================================== #
# TITLE                                                                #
# AMR: An R Package for Working with Antimicrobial Resistance Data     #
#                                                                      #
# SOURCE                                                               #
# https://github.com/msberends/AMR                                     #
#                                                                      #
# CITE AS                                                              #
# Berends MS, Luz CF, Friedrich AW, Sinha BNM, Albers CJ, Glasner C    #
# (2022). AMR: An R Package for Working with Antimicrobial Resistance  #
# Data. Journal of Statistical Software, 104(3), 1-31.                 #
# doi:10.18637/jss.v104.i03                                            #
#                                                                      #
# Developed at the University of Groningen, the Netherlands, in        #
# collaboration with non-profit organisations Certe Medical            #
# Diagnostics & Advice, and University Medical Center Groningen.       #
#                                                                      #
# 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 the full manual and a complete tutorial about  #
# how to conduct AMR data analysis: https://msberends.github.io/AMR/   #
# ==================================================================== #

# This script runs in under a minute and renews all guidelines of CLSI and EUCAST!
# Run it with source("data-raw/reproduction_of_rsi_translation.R")

library(dplyr)
library(readr)
library(tidyr)
library(AMR)

# Install the WHONET software on Windows (http://www.whonet.org/software.html),
# and copy the folder C:\WHONET\Codes to data-raw/WHONET/Codes
DRGLST <- read_tsv("data-raw/WHONET/Codes/DRGLST.txt", na = c("", "NA", "-"), show_col_types = FALSE)
DRGLST1 <- read_tsv("data-raw/WHONET/Codes/DRGLST1.txt", na = c("", "NA", "-"), show_col_types = FALSE)
ORGLIST <- read_tsv("data-raw/WHONET/Codes/ORGLIST.txt", na = c("", "NA", "-"), show_col_types = FALSE)

# create data set for generic rules (i.e., AB-specific but not MO-specific)
rsi_generic <- DRGLST %>%
  filter(CLSI == "X" | EUCST == "X") %>%
  select(ab = ANTIBIOTIC, disk_dose = POTENCY, matches("^(CLSI|EUCST)[0-9]")) %>%
  mutate(
    ab = as.ab(ab),
    across(matches("(CLSI|EUCST)"), as.double)
  ) %>%
  pivot_longer(-c(ab, disk_dose), names_to = "method") %>%
  separate(method, into = c("guideline", "method"), sep = "_") %>%
  mutate(method = ifelse(method %like% "D",
    gsub("D", "DISK_", method, fixed = TRUE),
    gsub("M", "MIC_", method, fixed = TRUE)
  )) %>%
  separate(method, into = c("method", "rsi"), sep = "_") %>%
  # I is in the middle, so we only need R and S (saves data)
  filter(rsi %in% c("R", "S")) %>%
  pivot_wider(names_from = rsi, values_from = value) %>%
  transmute(
    guideline = gsub("([0-9]+)$", " 20\\1", gsub("EUCST", "EUCAST", guideline)),
    method,
    site = NA_character_,
    mo = as.mo("UNKNOWN"),
    ab,
    ref_tbl = "Generic rules",
    disk_dose,
    breakpoint_S = S,
    breakpoint_R = R,
    uti = FALSE
  ) %>%
  filter(!(is.na(breakpoint_S) & is.na(breakpoint_R)), !is.na(mo), !is.na(ab))
rsi_generic

# create data set for AB-specific and MO-specific rules
rsi_specific <- DRGLST1 %>%
  # only support guidelines for humans (for now)
  filter(
    HOST == "Human" & SITE_INF %unlike% "canine|feline",
    # only CLSI and EUCAST
    GUIDELINES %like% "(CLSI|EUCST)"
  ) %>%
  # get microorganism names from another WHONET table
  mutate(ORG_CODE = tolower(ORG_CODE)) %>%
  left_join(ORGLIST %>%
    transmute(
      ORG_CODE = tolower(ORG),
      SCT_TEXT = case_when(
        is.na(SCT_TEXT) & is.na(ORGANISM) ~ ORG_CODE,
        is.na(SCT_TEXT) ~ ORGANISM,
        TRUE ~ SCT_TEXT
      )
    ) %>%
    # WHO for 'Generic'
    bind_rows(tibble(ORG_CODE = "gen", SCT_TEXT = "Unknown")) %>%
    # WHO for 'Enterobacterales'
    bind_rows(tibble(ORG_CODE = "ebc", SCT_TEXT = "Enterobacterales"))) %>%
  # still some manual cleaning required
  filter(!SCT_TEXT %in% c("Anaerobic Actinomycetes")) %>%
  transmute(
    guideline = gsub("([0-9]+)$", " 20\\1", gsub("EUCST", "EUCAST", GUIDELINES)),
    method = toupper(TESTMETHOD),
    site = SITE_INF,
    mo = as.mo(SCT_TEXT),
    ab = as.ab(WHON5_CODE),
    ref_tbl = REF_TABLE,
    disk_dose = POTENCY,
    breakpoint_S = as.double(ifelse(method == "DISK", DISK_S, MIC_S)),
    breakpoint_R = as.double(ifelse(method == "DISK", DISK_R, MIC_R)),
    uti = site %like% "(UTI|urinary|urine)"
  ) %>%
  filter(!(is.na(breakpoint_S) & is.na(breakpoint_R)), !is.na(mo), !is.na(ab))
rsi_specific

rsi_translation <- rsi_generic %>%
  bind_rows(rsi_specific) %>%
  # add the taxonomic rank index, used for sorting (so subspecies match first, order matches last)
  mutate(
    rank_index = case_when(
      mo_rank(mo) %like% "(infra|sub)" ~ 1,
      mo_rank(mo) == "species" ~ 2,
      mo_rank(mo) == "genus" ~ 3,
      mo_rank(mo) == "family" ~ 4,
      mo_rank(mo) == "order" ~ 5,
      TRUE ~ 6
    ),
    .after = mo
  ) %>%
  arrange(desc(guideline), ab, mo, method) %>%
  distinct(guideline, ab, mo, method, site, .keep_all = TRUE) %>%
  as.data.frame(stringsAsFactors = FALSE)

# disks MUST be 6-50 mm, so correct where that is wrong:
rsi_translation[which(rsi_translation$method == "DISK" &
  (is.na(rsi_translation$breakpoint_S) | rsi_translation$breakpoint_S > 50)), "breakpoint_S"] <- 50
rsi_translation[which(rsi_translation$method == "DISK" &
  (is.na(rsi_translation$breakpoint_R) | rsi_translation$breakpoint_R < 6)), "breakpoint_R"] <- 6
m <- unique(as.double(as.mic(levels(as.mic(1)))))
rsi_translation[which(rsi_translation$method == "MIC" &
  is.na(rsi_translation$breakpoint_S)), "breakpoint_S"] <- min(m)
rsi_translation[which(rsi_translation$method == "MIC" &
  is.na(rsi_translation$breakpoint_R)), "breakpoint_R"] <- max(m)

# WHONET has no >1024 but instead uses 1025, 513, etc, so raise these one higher valid MIC factor level:
rsi_translation[which(rsi_translation$breakpoint_R == 129), "breakpoint_R"] <- m[which(m == 128) + 1]
rsi_translation[which(rsi_translation$breakpoint_R == 257), "breakpoint_R"] <- m[which(m == 256) + 1]
rsi_translation[which(rsi_translation$breakpoint_R == 513), "breakpoint_R"] <- m[which(m == 512) + 1]
rsi_translation[which(rsi_translation$breakpoint_R == 1025), "breakpoint_R"] <- m[which(m == 1024) + 1]

# WHONET adds one log2 level to the R breakpoint for their software, e.g. in AMC in Enterobacterales:
# EUCAST 2021 guideline: S <= 8 and R > 8
#           WHONET file: S <= 8 and R >= 16
# this will make an MIC of 12 I, which should be R, so:
eucast_mics <- which(rsi_translation$guideline %like% "EUCAST" &
  rsi_translation$method == "MIC" &
  log2(as.double(rsi_translation$breakpoint_R)) - log2(as.double(rsi_translation$breakpoint_S)) != 0 &
  !is.na(rsi_translation$breakpoint_R))
old_R <- rsi_translation[eucast_mics, "breakpoint_R", drop = TRUE]
old_S <- rsi_translation[eucast_mics, "breakpoint_S", drop = TRUE]
new_R <- 2^(log2(old_R) - 1)
new_R[new_R < old_S | is.na(as.mic(new_R))] <- old_S[new_R < old_S | is.na(as.mic(new_R))]
rsi_translation[eucast_mics, "breakpoint_R"] <- new_R
eucast_disks <- which(rsi_translation$guideline %like% "EUCAST" &
  rsi_translation$method == "DISK" &
  rsi_translation$breakpoint_S - rsi_translation$breakpoint_R != 0 &
  !is.na(rsi_translation$breakpoint_R))
rsi_translation[eucast_disks, "breakpoint_R"] <- rsi_translation[eucast_disks, "breakpoint_R", drop = TRUE] + 1

# Greek symbols and EM dash symbols are not allowed by CRAN, so replace them with ASCII:
rsi_translation$disk_dose <- gsub("μ", "u", rsi_translation$disk_dose, fixed = TRUE)
rsi_translation$disk_dose <- gsub("–", "-", rsi_translation$disk_dose, fixed = TRUE)

# save to package
usethis::use_data(rsi_translation, overwrite = TRUE, compress = "xz")
rm(rsi_translation)
devtools::load_all(".")