AMR/R/mo_property.R

388 lines
15 KiB
R
Raw Normal View History

# ==================================================================== #
# TITLE #
# Antimicrobial Resistance (AMR) Analysis #
# #
2019-01-02 23:24:07 +01:00
# SOURCE #
# https://gitlab.com/msberends/AMR #
# #
# LICENCE #
2019-01-02 23:24:07 +01:00
# (c) 2019 Berends MS (m.s.berends@umcg.nl), Luz CF (c.f.luz@umcg.nl) #
# #
2019-01-02 23:24:07 +01:00
# 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. #
# #
# This R package was created for academic research and was publicly #
# released in the hope that it will be useful, but it comes WITHOUT #
# ANY WARRANTY OR LIABILITY. #
2019-04-05 18:47:39 +02:00
# Visit our website for more info: https://msberends.gitlab.io/AMR. #
# ==================================================================== #
#' Property of a microorganism
#'
#' Use these functions to return a specific property of a microorganism from the \code{\link{microorganisms}} data set. All input values will be evaluated internally with \code{\link{as.mo}}.
#' @param x any (vector of) text that can be coerced to a valid microorganism code with \code{\link{as.mo}}
#' @param property one of the column names of the \code{\link{microorganisms}} data set or \code{"shortname"}
2018-11-09 13:11:54 +01:00
#' @param language language of the returned text, defaults to system language (see \code{\link{get_locale}}) and can also be set with \code{\link{getOption}("AMR_locale")}. Use \code{language = NULL} or \code{language = ""} to prevent translation.
#' @param ... other parameters passed on to \code{\link{as.mo}}
2019-03-18 14:29:41 +01:00
#' @param open browse the URL using \code{\link[utils]{browseURL}()}
2019-02-20 00:04:48 +01:00
#' @details All functions will return the most recently known taxonomic property according to the Catalogue of Life, except for \code{mo_ref}, \code{mo_authors} and \code{mo_year}. This leads to the following results:
2018-11-09 13:11:54 +01:00
#' \itemize{
#' \item{\code{mo_name("Chlamydia psittaci")} will return \code{"Chlamydophila psittaci"} (with a warning about the renaming)}
2018-11-09 13:11:54 +01:00
#' \item{\code{mo_ref("Chlamydia psittaci")} will return \code{"Page, 1968"} (with a warning about the renaming)}
#' \item{\code{mo_ref("Chlamydophila psittaci")} will return \code{"Everett et al., 1999"} (without a warning)}
#' }
2019-02-20 00:04:48 +01:00
#'
2019-03-18 14:29:41 +01:00
#' The Gram stain - \code{mo_gramstain()} - will be determined on the taxonomic kingdom and phylum. According to Cavalier-Smith (2002) who defined subkingdoms Negibacteria and Posibacteria, only these phyla are Posibacteria: Actinobacteria, Chloroflexi, Firmicutes and Tenericutes. These bacteria are considered Gram positive - all other bacteria are considered Gram negative. Species outside the kingdom of Bacteria will return a value \code{NA}.
2019-02-20 00:04:48 +01:00
#'
2019-05-10 16:44:59 +02:00
#' All output will be \link{translate}d where possible.
#'
2019-03-18 14:29:41 +01:00
#' The function \code{mo_url()} will return the direct URL to the online database entry, which also shows the scientific reference of the concerned species.
2019-02-20 00:04:48 +01:00
#' @inheritSection catalogue_of_life Catalogue of Life
2018-09-24 23:33:29 +02:00
#' @inheritSection as.mo Source
#' @rdname mo_property
#' @name mo_property
2018-11-09 13:11:54 +01:00
#' @return \itemize{
2019-02-18 02:33:37 +01:00
#' \item{An \code{integer} in case of \code{mo_year}}
2018-11-09 13:11:54 +01:00
#' \item{A \code{list} in case of \code{mo_taxonomy}}
2019-02-22 22:12:10 +01:00
#' \item{A named \code{character} in case of \code{mo_url}}
2018-11-09 13:11:54 +01:00
#' \item{A \code{character} in all other cases}
#' }
#' @export
#' @seealso \code{\link{microorganisms}}
2019-01-02 23:24:07 +01:00
#' @inheritSection AMR Read more on our website!
#' @examples
2019-02-22 22:12:10 +01:00
#' ## taxonomic tree
2018-11-09 13:11:54 +01:00
#' mo_kingdom("E. coli") # "Bacteria"
2018-09-17 20:53:32 +02:00
#' mo_phylum("E. coli") # "Proteobacteria"
#' mo_class("E. coli") # "Gammaproteobacteria"
#' mo_order("E. coli") # "Enterobacteriales"
#' mo_family("E. coli") # "Enterobacteriaceae"
#' mo_genus("E. coli") # "Escherichia"
#' mo_species("E. coli") # "coli"
2019-02-20 00:04:48 +01:00
#' mo_subspecies("E. coli") # ""
2018-11-09 13:11:54 +01:00
#'
#' ## colloquial properties
2019-05-13 12:21:57 +02:00
#' mo_name("E. coli") # "Escherichia coli"
#' mo_fullname("E. coli") # "Escherichia coli", same as mo_name()
2018-09-05 10:51:46 +02:00
#' mo_shortname("E. coli") # "E. coli"
2018-11-09 13:11:54 +01:00
#'
#' ## other properties
2018-09-24 23:33:29 +02:00
#' mo_gramstain("E. coli") # "Gram negative"
2018-11-09 13:11:54 +01:00
#' mo_type("E. coli") # "Bacteria" (equal to kingdom)
2019-02-22 22:12:10 +01:00
#' mo_rank("E. coli") # "species"
2019-05-10 16:44:59 +02:00
#' mo_url("E. coli") # get the direct url to the online database entry
2018-11-09 13:11:54 +01:00
#'
#' ## scientific reference
2019-02-22 22:12:10 +01:00
#' mo_ref("E. coli") # "Castellani et al., 1919"
#' mo_authors("E. coli") # "Castellani et al."
2018-11-09 13:11:54 +01:00
#' mo_year("E. coli") # 1919
#'
#'
#' # Abbreviations known in the field
#' mo_genus("MRSA") # "Staphylococcus"
#' mo_species("MRSA") # "aureus"
2018-09-05 10:51:46 +02:00
#' mo_shortname("MRSA") # "S. aureus"
2018-09-24 23:33:29 +02:00
#' mo_gramstain("MRSA") # "Gram positive"
#'
#' mo_genus("VISA") # "Staphylococcus"
#' mo_species("VISA") # "aureus"
#'
#'
#' # Known subspecies
#' mo_name("doylei") # "Campylobacter jejuni doylei"
#' mo_genus("doylei") # "Campylobacter"
#' mo_species("doylei") # "jejuni"
#' mo_subspecies("doylei") # "doylei"
#'
2018-09-24 23:33:29 +02:00
#' mo_fullname("K. pneu rh") # "Klebsiella pneumoniae rhinoscleromatis"
2018-09-05 10:51:46 +02:00
#' mo_shortname("K. pneu rh") # "K. pneumoniae"
#'
#'
#' # Becker classification, see ?as.mo
2018-09-05 12:21:27 +02:00
#' mo_fullname("S. epi") # "Staphylococcus epidermidis"
2019-03-18 14:29:41 +01:00
#' mo_fullname("S. epi", Becker = TRUE) # "Coagulase-negative Staphylococcus (CoNS)"
2018-09-05 12:21:27 +02:00
#' mo_shortname("S. epi") # "S. epidermidis"
#' mo_shortname("S. epi", Becker = TRUE) # "CoNS"
#'
#' # Lancefield classification, see ?as.mo
2018-09-05 12:21:27 +02:00
#' mo_fullname("S. pyo") # "Streptococcus pyogenes"
#' mo_fullname("S. pyo", Lancefield = TRUE) # "Streptococcus group A"
#' mo_shortname("S. pyo") # "S. pyogenes"
2018-11-09 13:11:54 +01:00
#' mo_shortname("S. pyo", Lancefield = TRUE) # "GAS" ('Group A streptococci')
#'
#'
2019-02-22 22:12:10 +01:00
#' # language support for German, Dutch, Spanish, Portuguese, Italian and French
2018-09-24 23:33:29 +02:00
#' mo_gramstain("E. coli", language = "de") # "Gramnegativ"
#' mo_gramstain("E. coli", language = "nl") # "Gram-negatief"
#' mo_gramstain("E. coli", language = "es") # "Gram negativo"
#'
2018-11-09 13:11:54 +01:00
#' # mo_type is equal to mo_kingdom, but mo_kingdom will remain official
#' mo_kingdom("E. coli") # "Bacteria" on a German system
#' mo_type("E. coli") # "Bakterien" on a German system
#' mo_type("E. coli") # "Bacteria" on an English system
#'
2018-09-09 12:11:44 +02:00
#' mo_fullname("S. pyogenes",
#' Lancefield = TRUE,
#' language = "de") # "Streptococcus Gruppe A"
2018-09-09 12:11:44 +02:00
#' mo_fullname("S. pyogenes",
#' Lancefield = TRUE,
#' language = "nl") # "Streptococcus groep A"
2018-09-17 20:53:32 +02:00
#'
#'
2019-05-10 16:44:59 +02:00
#' # get a list with the complete taxonomy (from kingdom to subspecies)
2018-09-17 20:53:32 +02:00
#' mo_taxonomy("E. coli")
2019-05-13 12:21:57 +02:00
mo_name <- function(x, language = get_locale(), ...) {
mo_fullname(x = x, language = language, ... = ...)
}
#' @rdname mo_property
#' @export
2018-11-05 13:20:32 +01:00
mo_fullname <- function(x, language = get_locale(), ...) {
t(mo_validate(x = x, property = "fullname", ...), language = language)
}
2018-09-05 10:51:46 +02:00
#' @rdname mo_property
2019-04-05 18:47:39 +02:00
#' @importFrom dplyr %>% mutate pull
2018-09-05 10:51:46 +02:00
#' @export
2018-11-05 13:20:32 +01:00
mo_shortname <- function(x, language = get_locale(), ...) {
2018-10-01 11:39:43 +02:00
dots <- list(...)
Becker <- dots$Becker
if (is.null(Becker)) {
Becker <- FALSE
}
Lancefield <- dots$Lancefield
if (is.null(Lancefield)) {
Lancefield <- FALSE
}
2019-02-26 12:33:26 +01:00
2019-05-31 20:25:57 +02:00
# get result without transformations
res1 <- AMR::as.mo(x, Becker = FALSE, Lancefield = FALSE, reference_df = dots$reference_df)
# and result with transformations
res2 <- suppressWarnings(AMR::as.mo(res1, ...))
res2_fullname <- mo_fullname(res2, language = language)
res2_fullname[res2_fullname %like% " \\(CoNS\\)"] <- "CoNS"
res2_fullname[res2_fullname %like% " \\(CoPS\\)"] <- "CoPS"
res2_fullname[res2_fullname %like% " \\(KNS\\)"] <- "KNS"
res2_fullname[res2_fullname %like% " \\(KPS\\)"] <- "KPS"
res2_fullname[res2_fullname %like% " \\(CNS\\)"] <- "CNS"
res2_fullname[res2_fullname %like% " \\(CPS\\)"] <- "CPS"
res2_fullname[res2_fullname %like% " \\(SCN\\)"] <- "SCN"
res2_fullname <- gsub("Streptococcus (group|Gruppe|gruppe|groep|grupo|gruppo|groupe) (.)",
"G\\2S",
res2_fullname) # turn "Streptococcus group A" and "Streptococcus grupo A" to "GAS"
res2_fullname_vector <- res2_fullname[res2_fullname == mo_fullname(res1)]
res2_fullname[res2_fullname == mo_fullname(res1)] <- paste0(substr(mo_genus(res2_fullname_vector), 1, 1),
". ",
suppressWarnings(mo_species(res2_fullname_vector)))
if (sum(res1 == res2, na.rm = TRUE) > 0) {
res1[res1 == res2] <- paste0(substr(mo_genus(res1[res1 == res2]), 1, 1),
". ",
suppressWarnings(mo_species(res1[res1 == res2])))
2018-09-05 10:51:46 +02:00
}
2019-05-31 20:25:57 +02:00
res1[res1 != res2] <- res2_fullname
result <- as.character(res1)
2019-02-26 12:33:26 +01:00
2019-05-10 16:44:59 +02:00
t(result, language = language)
2018-09-05 10:51:46 +02:00
}
2018-09-17 20:53:32 +02:00
#' @rdname mo_property
#' @export
2018-11-05 13:20:32 +01:00
mo_subspecies <- function(x, language = get_locale(), ...) {
2019-05-10 16:44:59 +02:00
t(mo_validate(x = x, property = "subspecies", ...), language = language)
2018-09-17 20:53:32 +02:00
}
#' @rdname mo_property
#' @export
2018-11-05 13:20:32 +01:00
mo_species <- function(x, language = get_locale(), ...) {
2019-05-10 16:44:59 +02:00
t(mo_validate(x = x, property = "species", ...), language = language)
2018-09-17 20:53:32 +02:00
}
#' @rdname mo_property
#' @export
2018-11-05 13:20:32 +01:00
mo_genus <- function(x, language = get_locale(), ...) {
2019-05-10 16:44:59 +02:00
t(mo_validate(x = x, property = "genus", ...), language = language)
2018-09-17 20:53:32 +02:00
}
#' @rdname mo_property
#' @export
2019-03-09 08:21:00 +01:00
mo_family <- function(x, language = get_locale(), ...) {
2019-05-10 16:44:59 +02:00
t(mo_validate(x = x, property = "family", ...), language = language)
2018-09-17 20:53:32 +02:00
}
#' @rdname mo_property
#' @export
2019-03-09 08:21:00 +01:00
mo_order <- function(x, language = get_locale(), ...) {
2019-05-10 16:44:59 +02:00
t(mo_validate(x = x, property = "order", ...), language = language)
2018-09-17 20:53:32 +02:00
}
#' @rdname mo_property
#' @export
2019-03-09 08:21:00 +01:00
mo_class <- function(x, language = get_locale(), ...) {
2019-05-10 16:44:59 +02:00
t(mo_validate(x = x, property = "class", ...), language = language)
2018-09-17 20:53:32 +02:00
}
#' @rdname mo_property
#' @export
2019-03-09 08:21:00 +01:00
mo_phylum <- function(x, language = get_locale(), ...) {
2019-05-10 16:44:59 +02:00
t(mo_validate(x = x, property = "phylum", ...), language = language)
2018-09-17 20:53:32 +02:00
}
2018-09-05 10:51:46 +02:00
#' @rdname mo_property
#' @export
2019-03-09 08:21:00 +01:00
mo_kingdom <- function(x, language = get_locale(), ...) {
2019-04-09 10:34:40 +02:00
if (all(x %in% AMR::microorganisms$kingdom)) {
return(x)
}
2019-05-10 16:44:59 +02:00
x <- as.mo(x, ...)
2019-04-05 18:47:39 +02:00
kngdm <- mo_validate(x = x, property = "kingdom", ...)
if (language != "en") {
2019-05-10 16:44:59 +02:00
kngdm[x == "UNKNOWN"] <- t(kngdm[x == "UNKNOWN"], language = language)
2019-04-05 18:47:39 +02:00
}
kngdm
}
#' @rdname mo_property
#' @export
2018-11-05 13:20:32 +01:00
mo_type <- function(x, language = get_locale(), ...) {
2019-05-10 16:44:59 +02:00
t(mo_validate(x = x, property = "kingdom", ...), language = language)
2018-11-09 13:11:54 +01:00
}
#' @rdname mo_property
#' @export
mo_gramstain <- function(x, language = get_locale(), ...) {
2019-05-10 16:44:59 +02:00
x.mo <- as.mo(x, ...)
2019-04-05 18:47:39 +02:00
x.phylum <- mo_phylum(x.mo, language = "en")
2019-02-28 13:56:28 +01:00
x[x.phylum %in% c("Actinobacteria",
"Chloroflexi",
"Firmicutes",
"Tenericutes")] <- "Gram positive"
2019-02-18 02:33:37 +01:00
x[x != "Gram positive"] <- "Gram negative"
2019-04-05 18:47:39 +02:00
x[mo_kingdom(x.mo, language = "en") != "Bacteria"] <- NA_character_
2019-02-28 13:56:28 +01:00
x[x.mo == "B_GRAMP"] <- "Gram positive"
x[x.mo == "B_GRAMN"] <- "Gram negative"
2019-05-10 16:44:59 +02:00
t(x, language = language)
2018-10-01 11:39:43 +02:00
}
#' @rdname mo_property
#' @export
2018-11-09 13:11:54 +01:00
mo_ref <- function(x, ...) {
mo_validate(x = x, property = "ref", ...)
}
#' @rdname mo_property
#' @export
2018-11-09 13:11:54 +01:00
mo_authors <- function(x, ...) {
x <- mo_validate(x = x, property = "ref", ...)
2019-02-22 22:12:10 +01:00
# remove last 4 digits and presumably the comma and space that preceeds them
2018-11-09 13:11:54 +01:00
x[!is.na(x)] <- gsub(",? ?[0-9]{4}", "", x[!is.na(x)])
2019-02-22 22:12:10 +01:00
suppressWarnings(x)
2018-11-09 13:11:54 +01:00
}
2018-09-24 23:33:29 +02:00
2018-11-09 13:11:54 +01:00
#' @rdname mo_property
#' @export
mo_year <- function(x, ...) {
x <- mo_validate(x = x, property = "ref", ...)
# get last 4 digits
x[!is.na(x)] <- gsub(".*([0-9]{4})$", "\\1", x[!is.na(x)])
2019-02-22 22:12:10 +01:00
suppressWarnings(as.integer(x))
}
#' @rdname mo_property
#' @export
mo_rank <- function(x, ...) {
mo_validate(x = x, property = "rank", ...)
}
2018-09-17 20:53:32 +02:00
#' @rdname mo_property
#' @export
2019-03-09 08:21:00 +01:00
mo_taxonomy <- function(x, language = get_locale(), ...) {
2018-10-01 11:39:43 +02:00
x <- AMR::as.mo(x, ...)
2019-03-09 08:21:00 +01:00
base::list(kingdom = mo_kingdom(x, language = language),
phylum = mo_phylum(x, language = language),
class = mo_class(x, language = language),
order = mo_order(x, language = language),
family = mo_family(x, language = language),
genus = mo_genus(x, language = language),
species = mo_species(x, language = language),
subspecies = mo_subspecies(x, language = language))
2018-09-17 20:53:32 +02:00
}
2019-02-20 00:04:48 +01:00
#' @rdname mo_property
2019-02-22 22:12:10 +01:00
#' @importFrom utils browseURL
2019-03-18 14:29:41 +01:00
#' @importFrom dplyr %>% left_join select mutate case_when
2019-02-20 00:04:48 +01:00
#' @export
2019-02-22 22:12:10 +01:00
mo_url <- function(x, open = FALSE, ...) {
2019-03-18 14:29:41 +01:00
mo <- AMR::as.mo(x = x, ... = ...)
df <- data.frame(mo, stringsAsFactors = FALSE) %>%
left_join(select(AMR::microorganisms, mo, source, species_id), by = "mo") %>%
mutate(url = case_when(source == "CoL" ~
paste0(gsub("{year}", catalogue_of_life$year, catalogue_of_life$url_CoL, fixed = TRUE), "details/species/id/", species_id),
source == "DSMZ" ~
paste0(catalogue_of_life$url_DSMZ, "?bnu_no=", species_id, "#", species_id),
TRUE ~
NA_character_))
u <- df$url
names(u) <- mo_fullname(mo)
2019-02-22 22:12:10 +01:00
if (open == TRUE) {
if (length(u) > 1) {
2019-03-18 14:29:41 +01:00
warning("only the first URL will be opened, as `browseURL()` only suports one string.")
2019-02-22 22:12:10 +01:00
}
browseURL(u[1L])
}
2019-02-20 00:04:48 +01:00
u
}
2018-11-09 13:11:54 +01:00
#' @rdname mo_property
#' @importFrom data.table data.table as.data.table setkey
#' @export
mo_property <- function(x, property = 'fullname', language = get_locale(), ...) {
if (length(property) != 1L) {
stop("'property' must be of length 1.")
}
if (!property %in% colnames(AMR::microorganisms)) {
stop("invalid property: '", property, "' - use a column name of the `microorganisms` data set")
}
2019-05-10 16:44:59 +02:00
t(mo_validate(x = x, property = property, ...), language = language)
}
2018-10-01 11:39:43 +02:00
mo_validate <- function(x, property, ...) {
dots <- list(...)
Becker <- dots$Becker
if (is.null(Becker)) {
Becker <- FALSE
}
Lancefield <- dots$Lancefield
if (is.null(Lancefield)) {
Lancefield <- FALSE
}
2018-12-06 14:36:39 +01:00
if (!"AMR" %in% base::.packages()) {
2019-06-01 20:40:49 +02:00
require("AMR")
2019-02-18 02:33:37 +01:00
# check onLoad() in R/zzz.R: data tables are created there.
2018-12-06 14:36:39 +01:00
}
2019-03-26 14:24:03 +01:00
# try to catch an error when inputting an invalid parameter
2019-05-10 16:44:59 +02:00
# so the 'call.' can be set to FALSE
2019-03-26 14:24:03 +01:00
tryCatch(x[1L] %in% AMR::microorganisms[1, property],
error = function(e) stop(e$message, call. = FALSE))
2019-05-28 16:50:40 +02:00
if (!all(x %in% pull(AMR::microorganisms, property))
2018-10-31 12:10:49 +01:00
| Becker %in% c(TRUE, "all")
| Lancefield %in% c(TRUE, "all")) {
2018-10-01 11:39:43 +02:00
exec_as.mo(x, property = property, ...)
} else {
if (property == "mo") {
return(structure(x, class = "mo"))
} else {
return(x)
}
}
}