Highlighting individual axis labels in bold using ggplot2

Question

I want to highlight individual axis labels in bold. I am aware of this answer by @MrFlick but I can't figure out how to do this a) for more than one item, and b) whether it's possible to use the names of the labels instead of the number of the item in that list (or expression).

Here is an example dataset:

require(ggplot2)
require(dplyr)
set.seed(36)
xx<-data.frame(YEAR=rep(c("X","Y"), each=20),
               CLONE=rep(c("A","B","C","D","E"), each=4, 2),
               TREAT=rep(c("T1","T2","T3","C"), 10),
               VALUE=sample(c(1:10), 40, replace=T))

Then I am sorting my labels according to a particular factor combination which is then supposed to be maintained across multiple panels of a plot. See my previous question here.

clone_order <- xx %>% subset(TREAT == "C"  & YEAR == "X") %>%
  arrange(-VALUE) %>% select(CLONE) %>% unlist()    
xx <- xx %>% mutate(CLONE = factor(CLONE, levels = clone_order))

ggplot(xx, aes(x=CLONE, y=VALUE, fill=YEAR)) + 
  geom_bar(stat="identity", position="dodge") +
  facet_wrap(~TREAT)

Now I want to bold Clone A, B and E. I am sure this will work somehow but I cannot figure out how. Ideally, it would be great to know how to do this by a) using the number of the item in the list/expression, and b) by using the label, e.g. A, B and E.

Solution 1

Here's a generic method to create the emboldening vector:

colorado <- function(src, boulder) {
  if (!is.factor(src)) src <- factor(src)                   # make sure it's a factor
  src_levels <- levels(src)                                 # retrieve the levels in their order
  brave <- boulder %in% src_levels                          # make sure everything we want to make bold is actually in the factor levels
  if (all(brave)) {                                         # if so
    b_pos <- purrr::map_int(boulder, ~which(.==src_levels)) # then find out where they are
    b_vec <- rep("plain", length(src_levels))               # make'm all plain first
    b_vec[b_pos] <- "bold"                                  # make our targets bold
    b_vec                                                   # return the new vector
  } else {
    stop("All elements of 'boulder' must be in src")
  }
}

ggplot(xx, aes(x=CLONE, y=VALUE, fill=YEAR)) + 
  geom_bar(stat="identity", position="dodge") +
  facet_wrap(~TREAT) +
  theme(axis.text.x=element_text(face=colorado(xx$CLONE, c("A", "B", "E"))))

Solution 2

I'm not sure if you can map label characteristics by name, but it's definitely possible to do it by position with a call to theme:

ggplot(xx, aes(x=CLONE, y=VALUE, fill=YEAR)) + 
  geom_bar(stat="identity", position="dodge") +
  facet_wrap(~TREAT) +
  theme(axis.text.x = element_text(face = c('bold', 'bold', 'plain', 'plain', 'bold')))

Note that the listed font faces for axis.text.x are the same length as the labels of your x-axis (five elements). This produces:

Solution 3

You can create a named vector of expressions (that turn text to bold) in scale_x_discrete and use parse=TRUE to evaluate the expressions:

ggplot(xx, aes(x=CLONE, y=VALUE, fill=YEAR)) + 
    geom_bar(stat="identity", position="dodge") +
    facet_wrap(~TREAT) +
    scale_x_discrete(labels=c("A"=expression(bold(A)), "C"=expression(bold(C)),
                              "E"=expression(bold(E)), parse=TRUE))

You can probably create the vector of expressions programmatically, rather than typing it out, but the way to do that is escaping me right now.

Comments

[Stefan]

Thanks! Using axis.text.x is definitely an option but a bit cumbersome since I have 34 factor levels. That's why I was hoping to somehow just specify the ones I want to be bold by name and not by counting the position. I upvoted your answer but would like to keep it open a little longer. Maybe someone has a different idea.

[jdobres]

Playing with this a little more, it looks like element_text works only by position and ignores naming. One option for reducing typing would be judicious use of rep, for instance: c(rep('plain', 10), rep(c('bold', 'plain', each = 2)) would make the first 10 labels plain, the next 2 bold, and the next 2 plain.

[Stefan]

Wow, this is great! Thanks! One more question, would it also be possible, besides making it bold, to increase the font size of the bold label by say 2pt using size= in axis.text.x?

[hrbrmstr]

Aye. Most of the element_text() params are vectorized, but you'd need to copy and adapt this function to be a size changer vs a face changer.

[Stefan]

Thanks @eipi10! I like this one too as it is simple (i.e. upvote). However the answer by @hrbrmstr provides me with more control.

[eipi10]

Yes, I like his answer too.

[user2498193]

Trying to adapt this to alternatively top or bottom justify selected labels. I wrote code but it just produces a blank plot. Replaced "plain" with 1.0, and "bold" with 0.0 in your colorado' function. And in plot call: theme(axis.text.x=element_text(vjust=colorado2(xx$CLONE, c("A", "B", "E"))))'. However changes take up half of plot which I don't want. Any ideas how to scale the 0 or 1 within the label area of the plot @hrbrmstr ?

[hrbrmstr]

def a new question anony-user @user2498193

[user2498193]

ok no problem @hrbrmstr - new question is up: stackoverflow.com/questions/48286116/… Appreciate your help if you have time !

[jesseaam]

Exactly what I needed—thanks @jdobres! I used a logical vector to then create my "face" vector.