AMR/R/ggplot_rsi.R

424 lines
16 KiB
R
Raw Normal View History

2018-08-11 21:30:00 +02:00
# ==================================================================== #
# TITLE #
# Antimicrobial Resistance (AMR) Analysis #
# #
2019-01-02 23:24:07 +01:00
# SOURCE #
# https://gitlab.com/msberends/AMR #
2018-08-11 21:30:00 +02:00
# #
# 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) #
2018-08-11 21:30:00 +02:00
# #
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. #
2018-08-11 21:30:00 +02:00
# ==================================================================== #
2019-01-27 19:30:40 +01:00
#' AMR plots with \code{ggplot2}
2018-08-11 21:30:00 +02:00
#'
2019-05-31 20:25:57 +02:00
#' Use these functions to create bar plots for antimicrobial resistance analysis. All functions rely on internal \code{\link[ggplot2]{ggplot}2} functions.
2018-08-11 21:30:00 +02:00
#' @param data a \code{data.frame} with column(s) of class \code{"rsi"} (see \code{\link{as.rsi}})
2018-09-16 22:11:17 +02:00
#' @param position position adjustment of bars, either \code{"fill"} (default when \code{fun} is \code{\link{count_df}}), \code{"stack"} (default when \code{fun} is \code{\link{portion_df}}) or \code{"dodge"}
2019-06-13 14:28:46 +02:00
#' @param x variable to show on x axis, either \code{"antibiotic"} (default) or \code{"interpretation"} or a grouping variable
#' @param fill variable to categorise using the plots legend, either \code{"antibiotic"} (default) or \code{"interpretation"} or a grouping variable
#' @param breaks numeric vector of positions
#' @param limits numeric vector of length two providing limits of the scale, use \code{NA} to refer to the existing minimum or maximum
2019-06-13 14:28:46 +02:00
#' @param facet variable to split plots by, either \code{"interpretation"} (default) or \code{"antibiotic"} or a grouping variable
2019-05-13 10:10:16 +02:00
#' @inheritParams portion
2018-08-29 16:35:32 +02:00
#' @param nrow (when using \code{facet}) number of rows
2019-05-31 20:25:57 +02:00
#' @param colours a named vector with colours for the bars. The names must be one or more of: S, SI, I, IR, R or be \code{FALSE} to use default \code{ggplot2} colours.
2019-06-27 11:57:45 +02:00
#' @param datalabels show datalabels using \code{labels_rsi_count}
2018-09-16 22:11:17 +02:00
#' @param datalabels.size size of the datalabels
#' @param datalabels.colour colour of the datalabels
2019-05-31 20:25:57 +02:00
#' @param title text to show as title of the plot
#' @param subtitle text to show as subtitle of the plot
#' @param caption text to show as caption of the plot
#' @param x.title text to show as x axis description
#' @param y.title text to show as y axis description
2018-08-23 21:27:15 +02:00
#' @param ... other parameters passed on to \code{geom_rsi}
2019-05-31 20:25:57 +02:00
#' @details At default, the names of antibiotics will be shown on the plots using \code{\link{ab_name}}. This can be set with the \code{translate_ab} parameter. See \code{\link{count_df}}.
2018-08-11 21:30:00 +02:00
#'
#' \strong{The functions}\cr
2019-06-27 11:57:45 +02:00
#' \code{geom_rsi} will take any variable from the data that has an \code{rsi} class (created with \code{\link{as.rsi}}) using \code{\link{rsi_df}} and will plot bars with the percentage R, I and S. The default behaviour is to have the bars stacked and to have the different antibiotics on the x axis.
2018-08-11 21:30:00 +02:00
#'
2018-08-12 17:44:06 +02:00
#' \code{facet_rsi} creates 2d plots (at default based on S/I/R) using \code{\link[ggplot2]{facet_wrap}}.
2018-08-11 21:30:00 +02:00
#'
2018-09-17 20:53:32 +02:00
#' \code{scale_y_percent} transforms the y axis to a 0 to 100\% range using \code{\link[ggplot2]{scale_continuous}}.
2018-08-11 21:30:00 +02:00
#'
2019-05-31 20:25:57 +02:00
#' \code{scale_rsi_colours} sets colours to the bars: pastel blue for S, pastel turquoise for I and pastel red for R, using \code{\link[ggplot2]{scale_brewer}}.
2018-08-11 21:30:00 +02:00
#'
2018-09-17 20:53:32 +02:00
#' \code{theme_rsi} is a \code{ggplot \link[ggplot2]{theme}} with minimal distraction.
2018-08-11 21:30:00 +02:00
#'
2018-09-16 22:11:17 +02:00
#' \code{labels_rsi_count} print datalabels on the bars with percentage and amount of isolates using \code{\link[ggplot2]{geom_text}}
#'
2018-08-11 21:30:00 +02:00
#' \code{ggplot_rsi} is a wrapper around all above functions that uses data as first input. This makes it possible to use this function after a pipe (\code{\%>\%}). See Examples.
#' @rdname ggplot_rsi
2018-11-15 12:42:35 +01:00
#' @importFrom utils installed.packages
2018-08-11 21:30:00 +02:00
#' @export
2019-01-02 23:24:07 +01:00
#' @inheritSection AMR Read more on our website!
2018-08-11 21:30:00 +02:00
#' @examples
#' library(dplyr)
#' library(ggplot2)
#'
#' # get antimicrobial results for drugs against a UTI:
2019-05-10 16:44:59 +02:00
#' ggplot(septic_patients %>% select(AMX, NIT, FOS, TMP, CIP)) +
2018-08-11 21:30:00 +02:00
#' geom_rsi()
#'
2018-08-13 16:42:37 +02:00
#' # prettify the plot using some additional functions:
2019-05-31 20:25:57 +02:00
#' df <- septic_patients %>% select(AMX, NIT, FOS, TMP, CIP)
2018-08-11 21:30:00 +02:00
#' ggplot(df) +
2018-08-13 16:42:37 +02:00
#' geom_rsi() +
2018-08-11 21:30:00 +02:00
#' scale_y_percent() +
#' scale_rsi_colours() +
2018-09-16 22:11:17 +02:00
#' labels_rsi_count() +
2018-08-11 21:30:00 +02:00
#' theme_rsi()
#'
#' # or better yet, simplify this using the wrapper function - a single command:
#' septic_patients %>%
2019-05-10 16:44:59 +02:00
#' select(AMX, NIT, FOS, TMP, CIP) %>%
2018-08-11 21:30:00 +02:00
#' ggplot_rsi()
2018-08-22 00:02:26 +02:00
#'
2018-09-17 20:53:32 +02:00
#' # get only portions and no counts:
2018-08-22 00:02:26 +02:00
#' septic_patients %>%
2019-05-10 16:44:59 +02:00
#' select(AMX, NIT, FOS, TMP, CIP) %>%
2019-06-27 11:57:45 +02:00
#' ggplot_rsi(datalabels = FALSE)
#'
#' # add other ggplot2 parameters as you like:
#' septic_patients %>%
2019-05-10 16:44:59 +02:00
#' select(AMX, NIT, FOS, TMP, CIP) %>%
#' ggplot_rsi(width = 0.5,
#' colour = "black",
#' size = 1,
#' linetype = 2,
#' alpha = 0.25)
2018-12-15 22:40:07 +01:00
#'
2019-05-31 20:25:57 +02:00
#' septic_patients %>%
#' select(AMX) %>%
#' ggplot_rsi(colours = c(SI = "yellow"))
#'
2018-12-15 22:40:07 +01:00
#' # resistance of ciprofloxacine per age group
#' septic_patients %>%
#' mutate(first_isolate = first_isolate(.)) %>%
#' filter(first_isolate == TRUE,
#' mo == as.mo("E. coli")) %>%
#' # `age_group` is also a function of this package:
#' group_by(age_group = age_groups(age)) %>%
#' select(age_group,
2019-05-10 16:44:59 +02:00
#' CIP) %>%
2018-12-15 22:40:07 +01:00
#' ggplot_rsi(x = "age_group")
2018-08-13 16:42:37 +02:00
#' \donttest{
2018-08-29 16:39:28 +02:00
#'
#' # for colourblind mode, use divergent colours from the viridis package:
#' septic_patients %>%
2019-05-10 16:44:59 +02:00
#' select(AMX, NIT, FOS, TMP, CIP) %>%
2018-08-29 16:39:28 +02:00
#' ggplot_rsi() + scale_fill_viridis_d()
2019-05-31 20:25:57 +02:00
#' # a shorter version which also adjusts data label colours:
#' septic_patients %>%
#' select(AMX, NIT, FOS, TMP, CIP) %>%
#' ggplot_rsi(colours = FALSE)
2018-08-29 16:39:28 +02:00
#'
#'
2018-09-13 14:48:34 +02:00
#' # it also supports groups (don't forget to use the group var on `x` or `facet`):
2018-08-11 21:30:00 +02:00
#' septic_patients %>%
2019-05-10 16:44:59 +02:00
#' select(hospital_id, AMX, NIT, FOS, TMP, CIP) %>%
2018-08-13 16:42:37 +02:00
#' group_by(hospital_id) %>%
2019-05-31 20:25:57 +02:00
#' ggplot_rsi(x = "hospital_id",
2019-06-13 14:28:46 +02:00
#' facet = "antibiotic",
2019-05-31 20:25:57 +02:00
#' nrow = 1,
#' title = "AMR of Anti-UTI Drugs Per Hospital",
#' x.title = "Hospital",
#' datalabels = FALSE)
#'
2019-05-31 20:25:57 +02:00
#' # genuine analysis: check 3 most prevalent microorganisms
#' septic_patients %>%
2018-08-13 16:42:37 +02:00
#' # create new bacterial ID's, with all CoNS under the same group (Becker et al.)
2018-08-31 13:36:19 +02:00
#' mutate(mo = as.mo(mo, Becker = TRUE)) %>%
2018-09-13 14:48:34 +02:00
#' # filter on top three bacterial ID's
#' filter(mo %in% top_freq(freq(.$mo), 3)) %>%
2018-08-13 16:42:37 +02:00
#' # filter on first isolates
2019-05-31 20:25:57 +02:00
#' filter_first_isolate() %>%
2018-09-13 14:48:34 +02:00
#' # get short MO names (like "E. coli")
2019-05-31 20:25:57 +02:00
#' mutate(bug = mo_shortname(mo, Becker = TRUE)) %>%
2018-09-13 14:48:34 +02:00
#' # select this short name and some antiseptic drugs
2019-05-31 20:25:57 +02:00
#' select(bug, CXM, GEN, CIP) %>%
2018-08-13 16:42:37 +02:00
#' # group by MO
2019-05-31 20:25:57 +02:00
#' group_by(bug) %>%
2018-08-13 16:42:37 +02:00
#' # plot the thing, putting MOs on the facet
2019-06-13 14:28:46 +02:00
#' ggplot_rsi(x = "antibiotic",
2019-05-31 20:25:57 +02:00
#' facet = "bug",
2018-09-13 14:48:34 +02:00
#' translate_ab = FALSE,
2019-05-31 20:25:57 +02:00
#' nrow = 1,
#' title = "AMR of Top Three Microorganisms In Blood Culture Isolates",
2019-05-31 21:20:16 +02:00
#' subtitle = expression(paste("Only First Isolates, CoNS grouped according to Becker ",
#' italic("et al."), " (2014)")),
2019-05-31 20:25:57 +02:00
#' x.title = "Antibiotic (EARS-Net code)")
2018-08-13 16:42:37 +02:00
#' }
2018-08-11 21:30:00 +02:00
ggplot_rsi <- function(data,
2018-08-22 00:02:26 +02:00
position = NULL,
2019-06-13 14:28:46 +02:00
x = "antibiotic",
fill = "interpretation",
2018-08-23 21:27:15 +02:00
# params = list(),
2018-08-13 16:42:37 +02:00
facet = NULL,
breaks = seq(0, 1, 0.1),
limits = NULL,
2019-05-10 16:44:59 +02:00
translate_ab = "name",
2019-05-13 10:10:16 +02:00
combine_SI = TRUE,
combine_IR = FALSE,
2019-05-10 16:44:59 +02:00
language = get_locale(),
2018-08-29 16:35:32 +02:00
nrow = NULL,
2019-05-31 20:25:57 +02:00
colours = c(S = "#61a8ff",
SI = "#61a8ff",
I = "#61f7ff",
IR = "#ff6961",
R = "#ff6961"),
datalabels = TRUE,
datalabels.size = 2.5,
datalabels.colour = "gray15",
title = NULL,
subtitle = NULL,
caption = NULL,
2019-08-09 23:22:10 +02:00
x.title = "Antimicrobial",
y.title = "Proportion",
2018-08-13 16:42:37 +02:00
...) {
2018-08-12 17:44:06 +02:00
2019-03-26 14:24:03 +01:00
stopifnot_installed_package("ggplot2")
2018-08-12 17:44:06 +02:00
2018-09-13 14:48:34 +02:00
x <- x[1]
facet <- facet[1]
# we work with aes_string later on
x_deparse <- deparse(substitute(x))
if (x_deparse != "x") {
x <- x_deparse
}
if (x %like% '".*"') {
x <- substr(x, 2, nchar(x) - 1)
}
facet_deparse <- deparse(substitute(facet))
if (facet_deparse != "facet") {
facet <- facet_deparse
}
if (facet %like% '".*"') {
facet <- substr(facet, 2, nchar(facet) - 1)
}
if (facet %in% c("NULL", "")) {
facet <- NULL
}
2019-05-31 20:25:57 +02:00
if (is.null(position)) {
position <- "fill"
}
2018-08-11 21:30:00 +02:00
p <- ggplot2::ggplot(data = data) +
2019-05-13 10:10:16 +02:00
geom_rsi(position = position, x = x, fill = fill, translate_ab = translate_ab,
2019-06-27 11:57:45 +02:00
combine_SI = combine_SI, combine_IR = combine_IR, ...) +
2018-08-11 21:30:00 +02:00
theme_rsi()
2019-06-13 14:28:46 +02:00
if (fill == "interpretation") {
# set RSI colours
2019-05-31 20:25:57 +02:00
if (isFALSE(colours) & missing(datalabels.colour)) {
# set datalabel colour to middle gray
datalabels.colour <- "gray50"
}
p <- p + scale_rsi_colours(colours = colours)
2018-09-16 22:11:17 +02:00
}
2019-05-31 20:25:57 +02:00
2019-06-27 11:57:45 +02:00
if (identical(position, "fill")) {
2018-08-22 00:02:26 +02:00
# portions, so use y scale with percentage
p <- p + scale_y_percent(breaks = breaks, limits = limits)
2018-08-22 00:02:26 +02:00
}
2019-06-27 11:57:45 +02:00
if (datalabels == TRUE) {
2018-09-16 22:11:17 +02:00
p <- p + labels_rsi_count(position = position,
x = x,
2019-05-31 20:25:57 +02:00
translate_ab = translate_ab,
combine_SI = combine_SI,
combine_IR = combine_IR,
2018-09-16 22:11:17 +02:00
datalabels.size = datalabels.size,
datalabels.colour = datalabels.colour)
}
2018-08-11 21:30:00 +02:00
if (!is.null(facet)) {
2018-08-29 16:35:32 +02:00
p <- p + facet_rsi(facet = facet, nrow = nrow)
2018-08-11 21:30:00 +02:00
}
2019-05-31 20:25:57 +02:00
p <- p + ggplot2::labs(title = title,
subtitle = subtitle,
caption = caption,
x = x.title,
y = y.title)
2018-08-11 21:30:00 +02:00
p
}
#' @rdname ggplot_rsi
#' @export
2018-08-22 00:02:26 +02:00
geom_rsi <- function(position = NULL,
2019-06-13 14:28:46 +02:00
x = c("antibiotic", "interpretation"),
fill = "interpretation",
2019-05-10 16:44:59 +02:00
translate_ab = "name",
language = get_locale(),
2019-05-13 10:10:16 +02:00
combine_SI = TRUE,
combine_IR = FALSE,
2018-08-23 21:27:15 +02:00
...) {
2018-08-22 00:02:26 +02:00
2019-03-26 14:24:03 +01:00
stopifnot_installed_package("ggplot2")
2019-05-13 10:10:16 +02:00
if (is.data.frame(position)) {
stop("`position` is invalid. Did you accidentally use '%>%' instead of '+'?", call. = FALSE)
}
2019-06-13 14:28:46 +02:00
y <- "value"
2019-06-27 11:57:45 +02:00
if (missing(position) | is.null(position)) {
position <- "fill"
2018-08-22 00:02:26 +02:00
}
2018-08-11 21:30:00 +02:00
2019-05-31 20:25:57 +02:00
if (identical(position, "fill")) {
position <- ggplot2::position_fill(vjust = 0.5, reverse = TRUE)
}
2018-08-11 21:30:00 +02:00
x <- x[1]
2018-09-13 14:48:34 +02:00
# we work with aes_string later on
x_deparse <- deparse(substitute(x))
if (x_deparse != "x") {
x <- x_deparse
}
if (x %like% '".*"') {
x <- substr(x, 2, nchar(x) - 1)
}
2019-06-13 14:28:46 +02:00
if (tolower(x) %in% tolower(c('ab', 'abx', 'antibiotics'))) {
x <- "antibiotic"
} else if (tolower(x) %in% tolower(c('SIR', 'RSI', 'interpretations', 'result'))) {
x <- "interpretation"
2018-08-11 21:30:00 +02:00
}
ggplot2::layer(geom = "bar", stat = "identity", position = position,
2018-08-22 00:02:26 +02:00
mapping = ggplot2::aes_string(x = x, y = y, fill = fill),
2019-05-13 10:10:16 +02:00
params = list(...), data = function(x) {
2019-06-27 11:57:45 +02:00
AMR::rsi_df(data = x,
translate_ab = translate_ab,
language = language,
combine_SI = combine_SI,
combine_IR = combine_IR)
2019-05-13 10:10:16 +02:00
})
2018-08-11 21:30:00 +02:00
}
#' @rdname ggplot_rsi
#' @export
2019-06-13 14:28:46 +02:00
facet_rsi <- function(facet = c("interpretation", "antibiotic"), nrow = NULL) {
2018-08-13 16:42:37 +02:00
2019-03-26 14:24:03 +01:00
stopifnot_installed_package("ggplot2")
2019-05-31 20:25:57 +02:00
facet <- facet[1]
2018-09-13 14:48:34 +02:00
# we work with aes_string later on
facet_deparse <- deparse(substitute(facet))
if (facet_deparse != "facet") {
facet <- facet_deparse
}
if (facet %like% '".*"') {
facet <- substr(facet, 2, nchar(facet) - 1)
}
2019-06-13 14:28:46 +02:00
if (tolower(facet) %in% tolower(c('SIR', 'RSI', 'interpretations', 'result'))) {
facet <- "interpretation"
} else if (tolower(facet) %in% tolower(c('ab', 'abx', 'antibiotics'))) {
facet <- "antibiotic"
2018-08-11 21:30:00 +02:00
}
2018-08-13 16:42:37 +02:00
2018-08-29 16:35:32 +02:00
ggplot2::facet_wrap(facets = facet, scales = "free_x", nrow = nrow)
2018-08-11 21:30:00 +02:00
}
#' @rdname ggplot_rsi
#' @export
scale_y_percent <- function(breaks = seq(0, 1, 0.1), limits = NULL) {
2019-03-26 14:24:03 +01:00
stopifnot_installed_package("ggplot2")
2019-01-02 23:24:07 +01:00
if (all(breaks[breaks != 0] > 1)) {
breaks <- breaks / 100
}
ggplot2::scale_y_continuous(breaks = breaks,
labels = percent(breaks),
limits = limits)
2018-08-11 21:30:00 +02:00
}
#' @rdname ggplot_rsi
#' @export
2019-05-31 20:25:57 +02:00
scale_rsi_colours <- function(colours = c(S = "#61a8ff",
SI = "#61a8ff",
I = "#61f7ff",
IR = "#ff6961",
R = "#ff6961")) {
2019-03-26 14:24:03 +01:00
stopifnot_installed_package("ggplot2")
2019-01-29 00:06:50 +01:00
#ggplot2::scale_fill_brewer(palette = "RdYlGn")
2019-05-13 10:10:16 +02:00
#ggplot2::scale_fill_manual(values = c("#b22222", "#ae9c20", "#7cfc00"))
2019-05-31 20:25:57 +02:00
if (!identical(colours, FALSE)) {
original_cols <- c(S = "#61a8ff",
SI = "#61a8ff",
I = "#61f7ff",
IR = "#ff6961",
R = "#ff6961")
colours <- replace(original_cols, names(colours), colours)
ggplot2::scale_fill_manual(values = colours)
}
2018-08-11 21:30:00 +02:00
}
#' @rdname ggplot_rsi
#' @export
theme_rsi <- function() {
2019-03-26 14:24:03 +01:00
stopifnot_installed_package("ggplot2")
2019-05-31 20:25:57 +02:00
ggplot2::theme_minimal(base_size = 10) +
2018-08-22 00:02:26 +02:00
ggplot2::theme(panel.grid.major.x = ggplot2::element_blank(),
panel.grid.minor = ggplot2::element_blank(),
2019-05-31 20:25:57 +02:00
panel.grid.major.y = ggplot2::element_line(colour = "grey75"),
# center title and subtitle
plot.title = ggplot2::element_text(hjust = 0.5),
plot.subtitle = ggplot2::element_text(hjust = 0.5))
2018-08-11 21:30:00 +02:00
}
2018-09-16 22:11:17 +02:00
#' @rdname ggplot_rsi
2019-05-31 20:25:57 +02:00
#' @importFrom dplyr mutate %>% group_by_at
2018-09-16 22:11:17 +02:00
#' @export
labels_rsi_count <- function(position = NULL,
2019-06-13 14:28:46 +02:00
x = "antibiotic",
2019-05-31 20:25:57 +02:00
translate_ab = "name",
combine_SI = TRUE,
combine_IR = FALSE,
2018-09-16 22:11:17 +02:00
datalabels.size = 3,
2019-05-31 20:25:57 +02:00
datalabels.colour = "gray15") {
2019-03-26 14:24:03 +01:00
stopifnot_installed_package("ggplot2")
2018-09-16 22:11:17 +02:00
if (is.null(position)) {
position <- "fill"
}
2019-05-30 08:51:38 +02:00
if (identical(position, "fill")) {
position <- ggplot2::position_fill(vjust = 0.5, reverse = TRUE)
2018-09-16 22:11:17 +02:00
}
2019-05-31 20:25:57 +02:00
x_name <- x
2018-09-16 22:11:17 +02:00
ggplot2::geom_text(mapping = ggplot2::aes_string(label = "lbl",
x = x,
2019-06-13 14:28:46 +02:00
y = "value"),
2018-09-16 22:11:17 +02:00
position = position,
inherit.aes = FALSE,
size = datalabels.size,
2019-05-31 20:25:57 +02:00
colour = datalabels.colour,
lineheight = 0.75,
data = function(x) {
2019-06-27 11:57:45 +02:00
rsi_df(data = x,
translate_ab = translate_ab,
combine_SI = combine_SI,
combine_IR = combine_IR) %>%
2019-05-31 20:25:57 +02:00
group_by_at(x_name) %>%
2019-06-13 14:28:46 +02:00
mutate(lbl = paste0(percent(value / sum(value, na.rm = TRUE), force_zero = TRUE),
2019-06-27 11:57:45 +02:00
"\n(n=", isolates, ")"))
2019-05-31 20:25:57 +02:00
})
2018-09-16 22:11:17 +02:00
}