2018-08-11 21:30:00 +02:00
# ==================================================================== #
# TITLE #
# Antimicrobial Resistance (AMR) Analysis #
# #
# AUTHORS #
# Berends MS (m.s.berends@umcg.nl), Luz CF (c.f.luz@umcg.nl) #
# #
# LICENCE #
# This program is free software; you can redistribute it and/or modify #
# it under the terms of the GNU General Public License version 2.0, #
# as published by the Free Software Foundation. #
# #
# This program is distributed in the hope that it will be useful, #
# but WITHOUT ANY WARRANTY; without even the implied warranty of #
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the #
# GNU General Public License for more details. #
# ==================================================================== #
#' AMR bar plots with \code{ggplot}
#'
2018-08-12 17:44:06 +02:00
#' Use these functions to create bar plots for antimicrobial resistance analysis. All functions rely on internal \code{\link[ggplot2]{ggplot}} 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"}
2018-08-13 16:42:37 +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 facet variable to split plots by, either \code{"Interpretation"} (default) or \code{"Antibiotic"} or a grouping variable
#' @param translate_ab a column name of the \code{\link{antibiotics}} data set to translate the antibiotic abbreviations into, using \code{\link{abname}}. Default behaviour is to translate to official names according to the WHO. Use \code{translate_ab = FALSE} to disable translation.
2018-09-16 22:11:17 +02:00
#' @param fun function to transform \code{data}, either \code{\link{count_df}} (default) or \code{\link{portion_df}}
2018-08-29 16:35:32 +02:00
#' @param nrow (when using \code{facet}) number of rows
2018-09-16 22:11:17 +02:00
#' @param datalabels show datalabels using \code{labels_rsi_count}, will at default only be shown when \code{fun = count_df}
#' @param datalabels.size size of the datalabels
#' @param datalabels.colour colour of the datalabels
2018-08-23 21:27:15 +02:00
#' @param ... other parameters passed on to \code{geom_rsi}
2018-08-11 21:30:00 +02:00
#' @details At default, the names of antibiotics will be shown on the plots using \code{\link{abname}}. This can be set with the option \code{get_antibiotic_names} (a logical value), so change it e.g. to \code{FALSE} with \code{options(get_antibiotic_names = FALSE)}.
#'
#' \strong{The functions}\cr
2018-09-17 20:53:32 +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{fun} (\code{\link{count_df}} at default, can also be \code{\link{portion_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
#'
2018-10-10 10:50:19 +02:00
#' \code{scale_rsi_colours} sets colours to the bars: green for S, yellow for I and 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
#' @export
#' @examples
#' library(dplyr)
#' library(ggplot2)
#'
#' # get antimicrobial results for drugs against a UTI:
#' ggplot(septic_patients %>% select(amox, nitr, fosf, trim, cipr)) +
#' geom_rsi()
#'
2018-08-13 16:42:37 +02:00
#' # prettify the plot using some additional functions:
2018-08-11 21:30:00 +02:00
#' df <- septic_patients[, c("amox", "nitr", "fosf", "trim", "cipr")]
#' 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 %>%
#' select(amox, nitr, fosf, trim, cipr) %>%
#' 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 %>%
#' select(amox, nitr, fosf, trim, cipr) %>%
2018-09-17 20:53:32 +02:00
#' ggplot_rsi(fun = portion_df)
2018-08-29 16:25:57 +02:00
#'
#' # add other ggplot2 parameters as you like:
#' septic_patients %>%
#' select(amox, nitr, fosf, trim, cipr) %>%
#' ggplot_rsi(width = 0.5,
#' colour = "black",
#' size = 1,
#' linetype = 2,
#' alpha = 0.25)
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 %>%
#' select(amox, nitr, fosf, trim, cipr) %>%
#' ggplot_rsi() + scale_fill_viridis_d()
#'
#'
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 %>%
2018-08-13 16:42:37 +02:00
#' select(hospital_id, amox, nitr, fosf, trim, cipr) %>%
#' group_by(hospital_id) %>%
2018-09-13 14:48:34 +02:00
#' ggplot_rsi(x = hospital_id,
#' facet = Antibiotic,
2018-08-13 16:42:37 +02:00
#' nrow = 1) +
#' labs(title = "AMR of Anti-UTI Drugs Per Hospital",
#' x = "Hospital")
2018-08-12 22:34:03 +02:00
#'
2018-08-13 16:42:37 +02:00
#' # genuine analysis: check 2 most prevalent microorganisms
2018-08-12 22:34:03 +02:00
#' 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
#' # determine first isolates
#' mutate(first_isolate = first_isolate(.,
#' col_date = "date",
#' col_patient_id = "patient_id",
2018-08-31 13:36:19 +02:00
#' col_mo = "mo")) %>%
2018-08-13 16:42:37 +02:00
#' # filter on first isolates
#' filter(first_isolate == TRUE) %>%
2018-09-13 14:48:34 +02:00
#' # get short MO names (like "E. coli")
#' mutate(mo = mo_shortname(mo, Becker = TRUE)) %>%
#' # select this short name and some antiseptic drugs
#' select(mo, cfur, gent, cipr) %>%
2018-08-13 16:42:37 +02:00
#' # group by MO
#' group_by(mo) %>%
#' # plot the thing, putting MOs on the facet
2018-09-13 14:48:34 +02:00
#' ggplot_rsi(x = Antibiotic,
#' facet = mo,
#' translate_ab = FALSE,
#' nrow = 1) +
#' labs(title = "AMR of Top Three Microorganisms In Blood Culture Isolates",
2018-08-31 13:36:19 +02:00
#' subtitle = "Only First Isolates, CoNS grouped according to Becker et al. (2014)",
2018-08-13 16:42:37 +02:00
#' x = "Microorganisms")
#' }
2018-08-11 21:30:00 +02:00
ggplot_rsi <- function ( data ,
2018-08-22 00:02:26 +02:00
position = NULL ,
2018-08-11 21:30:00 +02:00
x = " Antibiotic" ,
2018-08-12 22:34:03 +02:00
fill = " Interpretation" ,
2018-08-23 21:27:15 +02:00
# params = list(),
2018-08-13 16:42:37 +02:00
facet = NULL ,
translate_ab = " official" ,
2018-09-16 22:11:17 +02:00
fun = count_df ,
2018-08-29 16:35:32 +02:00
nrow = NULL ,
2018-09-16 22:11:17 +02:00
datalabels = TRUE ,
datalabels.size = 3 ,
datalabels.colour = " grey15" ,
2018-08-13 16:42:37 +02:00
... ) {
2018-08-12 17:44:06 +02:00
if ( ! " ggplot2" %in% rownames ( installed.packages ( ) ) ) {
stop ( ' this function requires the ggplot2 package.' , call. = FALSE )
}
2018-08-22 00:02:26 +02:00
fun_name <- deparse ( substitute ( fun ) )
if ( ! fun_name %in% c ( " portion_df" , " count_df" ) ) {
stop ( " `fun` must be portion_df or count_df" )
}
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
}
2018-08-11 21:30:00 +02:00
p <- ggplot2 :: ggplot ( data = data ) +
2018-08-23 21:27:15 +02:00
geom_rsi ( position = position , x = x , fill = fill , translate_ab = translate_ab , fun = fun , ... ) +
2018-08-11 21:30:00 +02:00
theme_rsi ( )
2018-08-12 22:34:03 +02:00
if ( fill == " Interpretation" ) {
# set RSI colours
p <- p + scale_rsi_colours ( )
}
2018-09-16 22:11:17 +02:00
if ( is.null ( position ) ) {
position <- " fill"
}
if ( fun_name == " portion_df"
| ( fun_name == " count_df" & position == " fill" ) ) {
2018-08-22 00:02:26 +02:00
# portions, so use y scale with percentage
p <- p + scale_y_percent ( )
}
2018-08-12 22:34:03 +02:00
2018-09-16 22:11:17 +02:00
if ( fun_name == " count_df" & datalabels == TRUE ) {
p <- p + labels_rsi_count ( position = position ,
x = x ,
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
}
p
}
#' @rdname ggplot_rsi
#' @export
2018-08-22 00:02:26 +02:00
geom_rsi <- function ( position = NULL ,
2018-08-13 16:42:37 +02:00
x = c ( " Antibiotic" , " Interpretation" ) ,
fill = " Interpretation" ,
2018-08-22 00:02:26 +02:00
translate_ab = " official" ,
2018-09-16 22:11:17 +02:00
fun = count_df ,
2018-08-23 21:27:15 +02:00
... ) {
2018-08-22 00:02:26 +02:00
fun_name <- deparse ( substitute ( fun ) )
if ( ! fun_name %in% c ( " portion_df" , " count_df" , " fun" ) ) {
stop ( " `fun` must be portion_df or count_df" )
}
2018-09-16 22:11:17 +02:00
y <- " Value"
2018-08-22 00:02:26 +02:00
if ( identical ( fun , count_df ) ) {
if ( missing ( position ) | is.null ( position ) ) {
2018-09-16 22:11:17 +02:00
position <- " fill"
2018-08-22 00:02:26 +02:00
}
} else {
if ( missing ( position ) | is.null ( position ) ) {
position <- " stack"
}
}
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 )
}
2018-08-23 21:27:15 +02:00
if ( tolower ( x ) %in% tolower ( c ( ' ab' , ' antibiotic' , ' abx' , ' antibiotics' ) ) ) {
2018-08-13 16:42:37 +02:00
x <- " Antibiotic"
2018-08-23 21:27:15 +02:00
} else if ( tolower ( x ) %in% tolower ( c ( ' SIR' , ' RSI' , ' interpretation' , ' interpretations' , ' result' ) ) ) {
2018-08-13 16:42:37 +02:00
x <- " Interpretation"
2018-08-11 21:30:00 +02:00
}
2018-08-13 16:42:37 +02:00
options ( get_antibiotic_names = translate_ab )
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 ) ,
2018-08-23 21:27:15 +02:00
data = fun , params = list ( ... ) )
2018-08-11 21:30:00 +02:00
}
#' @rdname ggplot_rsi
#' @export
2018-08-29 16:35:32 +02:00
facet_rsi <- function ( facet = c ( " Interpretation" , " Antibiotic" ) , nrow = NULL ) {
2018-08-13 16:42:37 +02:00
2018-09-13 14:48:34 +02:00
facet <- facet [1 ]
# 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 )
}
2018-08-23 21:27:15 +02:00
if ( tolower ( facet ) %in% tolower ( c ( ' SIR' , ' RSI' , ' interpretation' , ' interpretations' , ' result' ) ) ) {
2018-08-13 16:42:37 +02:00
facet <- " Interpretation"
2018-08-23 21:27:15 +02:00
} else if ( tolower ( facet ) %in% tolower ( c ( ' ab' , ' antibiotic' , ' abx' , ' antibiotics' ) ) ) {
2018-08-13 16:42:37 +02:00
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 ( ) {
2018-08-22 00:02:26 +02:00
ggplot2 :: scale_y_continuous ( breaks = seq ( 0 , 1 , 0.1 ) ,
2018-08-11 21:30:00 +02:00
labels = percent ( seq ( 0 , 1 , 0.1 ) ) )
}
#' @rdname ggplot_rsi
#' @export
scale_rsi_colours <- function ( ) {
ggplot2 :: scale_fill_brewer ( palette = " RdYlGn" )
}
#' @rdname ggplot_rsi
#' @export
theme_rsi <- function ( ) {
2018-08-22 00:02:26 +02:00
ggplot2 :: theme_minimal ( ) +
ggplot2 :: theme ( panel.grid.major.x = ggplot2 :: element_blank ( ) ,
panel.grid.minor = ggplot2 :: element_blank ( ) ,
panel.grid.major.y = ggplot2 :: element_line ( colour = " grey75" ) )
2018-08-11 21:30:00 +02:00
}
2018-09-16 22:11:17 +02:00
#' @rdname ggplot_rsi
#' @export
labels_rsi_count <- function ( position = NULL ,
x = " Antibiotic" ,
datalabels.size = 3 ,
datalabels.colour = " grey15" ) {
if ( is.null ( position ) ) {
position <- " fill"
}
if ( position == " fill" ) {
position <- ggplot2 :: position_fill ( vjust = 0.5 )
}
ggplot2 :: geom_text ( mapping = ggplot2 :: aes_string ( label = " lbl" ,
x = x ,
y = " Value" ) ,
position = position ,
data = getlbls ,
inherit.aes = FALSE ,
size = datalabels.size ,
colour = datalabels.colour )
}
#' @importFrom dplyr %>% group_by mutate
getlbls <- function ( data ) {
data %>%
count_df ( ) %>%
group_by ( Antibiotic ) %>%
mutate ( lbl = paste0 ( percent ( Value / sum ( Value , na.rm = TRUE ) , force_zero = TRUE ) ,
" (n=" , Value , " )" ) ) %>%
mutate ( lbl = ifelse ( lbl == " 0.0% (n=0)" , " " , lbl ) )
}