AMR/data-raw/reproduction_of_rsi_transla...

97 lines
3.5 KiB
R
Raw Normal View History

2019-05-10 16:44:59 +02:00
library(dplyr)
2020-07-29 10:33:47 +02:00
library(readr)
library(tidyr)
2019-05-10 16:44:59 +02:00
# Installed WHONET 2019 software on Windows (http://www.whonet.org/software.html),
# opened C:\WHONET\Codes\WHONETCodes.mdb in MS Access
# and exported table 'DRGLST1' to MS Excel
2020-02-14 19:54:13 +01:00
DRGLST1 <- readxl::read_excel("data-raw/DRGLST1.xlsx", na = c("", "NA", "-"))
2020-07-29 10:33:47 +02:00
rsi_trans <- DRGLST1 %>%
2019-05-10 16:44:59 +02:00
# only keep CLSI and EUCAST guidelines:
2020-07-29 10:33:47 +02:00
filter(GUIDELINES %like% "^(CLSI|EUCST)")
if (any(is.na(rsi_trans$BREAKPOINT_TYPE)) | !"Human" %in% rsi_trans$BREAKPOINT_TYPE) {
stop("Check column BREAKPOINT_TYPE - something is WRONG!")
}
rsi_trans <- rsi_trans %>%
filter(BREAKPOINT_TYPE == "Human") %>%
2020-01-27 11:37:00 +01:00
mutate(DISK_S = ifelse(as.double(DISK_S) > 50, 50, DISK_S),
MIC_R = ifelse(as.double(MIC_R) %in% c(1025, 129, 513), as.double(MIC_R) - 1, MIC_R)) %>%
2019-05-10 16:44:59 +02:00
# set a nice layout:
transmute(guideline = gsub("([0-9]+)$", " 20\\1", gsub("EUCST", "EUCAST", GUIDELINES)),
method = TESTMETHOD,
site = SITE_INF,
2019-05-10 16:44:59 +02:00
mo = as.mo(ORG_CODE),
ab = as.ab(WHON5_CODE),
ref_tbl = REF_TABLE,
dose_disk = POTENCY,
2020-01-27 11:37:00 +01:00
S_disk = as.disk(DISK_S),
2019-05-10 16:44:59 +02:00
R_disk = as.disk(DISK_R),
S_mic = as.mic(MIC_S),
2020-01-27 11:37:00 +01:00
R_mic = as.mic(MIC_R)) %>%
2020-02-20 13:19:23 +01:00
filter(!is.na(mo),
!is.na(ab),
!mo %in% c("UNKNOWN", "B_GRAMN", "B_GRAMP", "F_FUNGUS", "F_YEAST")) %>%
2019-05-10 16:44:59 +02:00
arrange(desc(guideline), mo, ab)
2019-09-18 15:46:09 +02:00
print(mo_failures())
2019-05-10 16:44:59 +02:00
# create 2 tables: MIC and disk
2020-07-29 10:33:47 +02:00
tbl_mic <- rsi_trans %>%
2019-05-10 16:44:59 +02:00
filter(method == "MIC") %>%
mutate(breakpoint_S = as.double(S_mic), breakpoint_R = as.double(R_mic))
2020-07-29 10:33:47 +02:00
tbl_disk <- rsi_trans %>%
2019-05-10 16:44:59 +02:00
filter(method == "DISK") %>%
mutate(breakpoint_S = as.double(S_disk), breakpoint_R = as.double(R_disk))
2019-05-10 16:44:59 +02:00
# merge them so every record is a unique combination of method, mo and ab
2020-07-29 10:33:47 +02:00
rsi_trans <- bind_rows(tbl_mic, tbl_disk) %>%
rename(disk_dose = dose_disk) %>%
mutate(disk_dose = gsub("µ", "u", disk_dose)) %>%
2020-02-14 19:54:13 +01:00
select(-ends_with("_mic"), -ends_with("_disk"))
2020-04-14 14:12:31 +02:00
# add extra CLSI general guidelines
clsi_general <- read_tsv("data-raw/DRGLST.txt") %>%
filter(CLSI == "X") %>%
select(WHON5_CODE,
disk_dose = POTENCY,
starts_with("CLSI"),
-c(CLSI, CLSI_ORDER)) %>%
mutate_at(vars(matches("CLSI")), as.double) %>%
pivot_longer(-c(WHON5_CODE, disk_dose)) %>%
mutate(method = ifelse(name %like% "_D", "DISK", "MIC"),
breakpoint = paste0("breakpoint_", gsub(".*([A-Z])$", "\\1", name)),
guideline = paste0("CLSI 20", cleaner::clean_integer(name))) %>%
filter(breakpoint != "breakpoint_I", !is.na(value)) %>%
select(-name) %>%
pivot_wider(names_from = breakpoint, values_from = value) %>%
transmute(guideline,
method,
site = NA_character_,
mo = as.mo("UNKNOWN"),
ab = as.ab(WHON5_CODE),
ref_tbl = "Generic CLSI rules",
disk_dose = gsub("/", "-", disk_dose, fixed = TRUE),
breakpoint_S,
breakpoint_R)
2020-02-14 19:54:13 +01:00
# add new EUCAST with read_EUCAST.R
2020-04-14 14:12:31 +02:00
# 2020-04-14 did that now for 2019 and 2020
2020-07-29 10:33:47 +02:00
rsi_trans <- rsi_trans %>%
filter(guideline != "EUCAST 2019") %>%
2020-02-14 19:54:13 +01:00
bind_rows(new_EUCAST) %>%
2020-04-14 14:12:31 +02:00
bind_rows(clsi_general) %>%
2020-02-14 19:54:13 +01:00
mutate(uti = site %like% "(UTI|urinary)") %>%
2019-05-10 16:44:59 +02:00
as.data.frame(stringsAsFactors = FALSE) %>%
# force classes again
mutate(mo = as.mo(mo),
2020-02-14 19:54:13 +01:00
ab = as.ab(ab)) %>%
arrange(desc(guideline), ab, mo, method)
2019-05-10 16:44:59 +02:00
# save to package
2020-07-29 10:33:47 +02:00
rsi_translation <- rsi_trans
2019-05-10 16:44:59 +02:00
usethis::use_data(rsi_translation, overwrite = TRUE)
2020-07-29 10:33:47 +02:00
rm(rsi_trans)
2019-05-10 16:44:59 +02:00
rm(rsi_translation)
2019-09-20 12:33:05 +02:00
devtools::load_all(".")