automatically adding brackets or braces to ggplot and plotly objects - r

Question

I want to automate adding brackets/braces to a ggplot object and then convert it to plotly using ggplotly.
library(tidyverse)
library(plotly)
#devtools::install_github("NicolasH2/ggbrace")
library(ggbrace)
set.seed(10)
mydata <- data.frame(xx = c(1:10), yy = sample(0:10, 10))
my_bracket_data <- data.frame(start = c(1, 4, 6),
end = c(3, 5, 9),
info = c("first", "second", "third"),
y_bottom = rep(11, 3),
y_top = rep(12, 3))
I can do this by using geom_brace but it involves manually typing it out for each brace rather than using the data that is already stored in a data.frame:
plot_geom_brace <- ggplot(data = mydata, aes(xx, yy)) +
geom_line(size = 1.5) +
geom_brace(aes(x = c(1, 3), y = c(11, 12), label = "first"), inherit.data = F, labelsize = 5) +
geom_brace(aes(x = c(4, 5), y = c(11, 12), label = "second"), inherit.data = F, labelsize = 5) +
geom_brace(aes(x = c(6, 9), y = c(11, 12), label = "third"), inherit.data = F, labelsize = 5)
plot_geom_brace
ggplotly(plot_geom_brace)
Is there a way that I can do this without repeatedly writing geom_brace layers for each brace (and instead access the data from my_bracket_data directly)?
As an aside this can be automated using geom_bracket but this is not supported by plotly yet.
library(ggpubr)
plot_geom_bracket <- ggplot(data = mydata, aes(xx, yy)) +
geom_line(size = 1.5) +
geom_bracket(xmin = my_bracket_data$start,
xmax = my_bracket_data$end,
y.position = rep(11, 3),
label = my_bracket_data$info,
inherit.aes = FALSE)
plot_geom_bracket
ggplotly(plot_geom_bracket)
# Warning message:
# In geom2trace.default(dots[[1L]][[1L]], dots[[2L]][[1L]], dots[[3L]][[1L]]) :
# geom_GeomBracket() has yet to be implemented in plotly.
# If you'd like to see this geom implemented,
# Please open an issue with your example code at
# https://github.com/ropensci/plotly/issues
Any suggestions?
Thanks

Answer 1

An option could be by creating two dataframes of your bracket data. One dataframe for the geom_braces by converting the data to a longer format with pivot_longer to create three geom braces through the aes. To get the labels you can create a small summarise table with the x and y positions per group of info. Here is some reproducible code:
library(tidyverse)
library(plotly)
#devtools::install_github("NicolasH2/ggbrace")
library(ggbrace)
set.seed(10)
mydata <- data.frame(xx = c(1:10), yy = sample(0:10, 10))
my_bracket_data <- data.frame(start = c(1, 4, 6),
end = c(3, 5, 9),
info = c("first", "second", "third"),
y_bottom = rep(11, 3),
y_top = rep(12, 3))
# Data for geom_brace
my_bracket_data_long <- my_bracket_data %>%
pivot_longer(cols = c(start, end), values_to = "x_value", names_to = "x_names") %>%
pivot_longer(cols = c(y_bottom, y_top), values_to = "y_value", names_to = "y_names")
# data for labels braces
my_bracket_data_labels <- my_bracket_data_long %>%
group_by(info) %>%
summarise(x_pos = mean(x_value),
y_pos = mean(y_value) + 1)
# plot
plot_geom_brace <- ggplot(data = mydata, aes(xx, yy)) +
geom_line(size = 1.5) +
geom_brace(data = my_bracket_data_long, aes(x = x_value, y = y_value, group = info)) +
geom_text(data = my_bracket_data_labels, aes(x = x_pos, y = y_pos, group = info, label = info))
ggplotly(plot_geom_brace)
Created on 2023-01-07 with reprex v2.0.2

Answer 2

special ggplot2 libraries like ggpubr usually don't play along well with conversion to plotly objects.
If you dont want to type out each geom_bracet call you could loop over the rows of the dataframe, create the geom statement using paste and pass it to the existing plot object using eval in the following line:
m<- data.frame(s = c(1, 4, 6), ## = my_bracket_data
e = c(3, 5, 9),
i = c("first", "second", "third"),
y_b = rep(11, 3),
y_t = rep(12, 3))
p<- ggplot(data = mydata, aes(xx, yy)) + geom_line(size = 1.5)
for (i in 1:NROW(my_bracket_data)) {
input = paste('geom_brace(aes(x =c(',m[i,]$s,',',m[i,]$e'),c(',m[i,]$y_b,',',
m[i,]$y_t,'),label=',m[i,]$i,'), inherit.data = F, labelsize = 5)',sep='')
p = p + eval(parse(text=input))
}
p
However this is more of a hacky solution, but that's what R tends to become if you incorporate/mix different styles like for example apply functions with tidyr syntax (or in this case ggplot, which could be seen as an ancestor of tidyr) and more programming style approaches (for, while , func...[yes you can programm in R]) and also want to let it automatically converse the whole thing to a Javascript thing (aka plotly) . .. its a beautiful mess .