ggplot: How to set default color for all geoms?

Question

I'm trying to set the default color for all geoms in a ggplot to something other than black. Note this is not about setting scale_color...

Simple example:

# linear model with confidence bands...
set.seed(1)
df <- data.frame(x=1:50, y=5 + 2*(1:50)+rnorm(50,sd=10))
lm <- lm(y~x,df)
se <- summary(lm)$sigma           # standard error of fit
Z  <- qnorm(0.05/2,lower.tail=F)  # 95% confidence bands
df <- cbind(df,predict(lm,se.fit=T)[c("fit","se.fit")])
# plot the result...
library(ggplot2)
ggplot(df, aes(x=x)) + 
  geom_point(aes(y=y), size=3) +
  geom_line(aes(y=fit)) +
  geom_line(aes(y=fit+Z*se.fit), linetype=2)+
  geom_line(aes(y=fit-Z*se.fit), linetype=2)

Now, suppose I want to make everything red. Leaving aside the advisability of doing that, I would think ggplot(df, aes(x=x), colour="red") would do it. But the colour= parameter seems to be ignored: everything is still black. I can add colour="red" to every geom_ call, but I'm trying to avoid that.

Edit: Using ggplot(df, aes(x=x, color="red")) is not an option because it creates a color scale using the default ggplot palette (evenly spaced around an HSL color circle). With one color, this is #F8766D, which happens to be light red. In addition, this creates a legend which then must be hidden.

Spoliate

Just curiously, why do you not want to use scale_color_? – Per

@Per The edit at the end of the question kind of explains it. I would have to create a color scale by setting a color in aes(...) in the ggplot call, then use something like scale_color_manual(values="red", guide="none"). It's just a hack - there should be an easier way. Plus, suppose I want to use a color scale for something else in the plot, say color the points based on some grouping variable and make everything else red. – Spoliate

@Per another reason would be for dark themed plots. There the only challenge is to set the geom related color (dots, box plots, lines etc.) – Constitute

Answer 1

You can set a default color for each geometry type this way:

update_geom_defaults("point",   list(colour = "red"))
update_geom_defaults("line",   list(colour = "red"))

ggplot(df, aes(x=x)) + 
  geom_point(aes(y=y), size=3) +
  geom_line(aes(y=fit)) +
  geom_line(aes(y=fit+Z*se.fit), linetype=2)+
  geom_line(aes(y=fit-Z*se.fit), linetype=2)

Edit If you want to do to everything then use (Edit borrow from here):

params <- ls(pattern = '^geom_', env = as.environment('package:ggplot2'))
geoms <- gsub("geom_", "", params)

lapply(geoms, update_geom_defaults, list(colour = "red"))
lapply(geoms, update_geom_defaults, list(fill = "red", colour = "red")) ## include fills 

If you want to set the default colour for the just one plot, simply do:

ggplot(df, aes(x=x, colour="red")) + 
  geom_point(aes(y=y), size=3) +
  geom_line(aes(y=fit)) +
  geom_line(aes(y=fit+Z*se.fit), linetype=2)+
  geom_line(aes(y=fit-Z*se.fit), linetype=2)

Living

Wow, I never knew about that function. Amazing. – Potbellied

Thanks but this is not quite what I was looking for. I'm trying to set the default color for just this ggplot object, in the same sense that calling ggplot(df) makes df the default dataset for this object. According to the documentation, update_geom_default(...) sets the defaults for all future plots in this session. – Spoliate

Amended to reflect that distinction. @Blue Magister, is the amendment what you suggested earlier? If so, please resubmit for attribution! – Living

@Living Amazing function! However, there is an issue with your code because it maps the colour aesthetic to a variable 'red', try changing it to 'green' and you'll see the lines and points are still colored red. – Interscholastic

Is there a way to do this with multiple colors, for instance with geom_col when there are 3 categories. – Quianaquibble

@amzu though aes(colour = "red") doesn't throw any error, technically it is wrong. It assigns default ggplot color to all data and not red color. The alternative is plot + scale_color_manual( breaks = NULL , values = "red") – Boater

Answer 2

In order to replace a geom default aesthetic with another one (for all geoms using that aesthetic), you can try the following code.

First define a function for getting default aes settings of all geoms from ggplot2

library(ggplot2)
library(purrr)

geom_aes_defaults <- function() {
  geom_names <- apropos("^Geom", ignore.case = FALSE)
  geoms <- mget(geom_names, env = asNamespace("ggplot2"))
  map(geoms, ~ .$default_aes)
}

With geom_aes_defaults() you obtain a long list of all geom aesthetic mappings

$Geom
Aesthetic mapping:
<empty>

$GeomAbline
Aesthetic mapping:
* `colour`   -> "black"
* `size`     -> 0.5
* `linetype` -> 1
* `alpha`    -> NA

$GeomAnnotationMap
Aesthetic mapping:
* `colour`   -> "NA"
* `fill`     -> "grey20"
* `size`     -> 0.5
* `linetype` -> 1
* `alpha`    -> NA

$GeomArea
Aesthetic mapping:
* `colour`   -> NA
* `fill`     -> "grey20"
* `size`     -> 0.5
* `linetype` -> 1
* `alpha`    -> NA

...

The following function iterates over all geoms matching a given aesthetic and substitutes the corresponding values

replace_geom_aes_defaults <- function(name, old_aes, new_aes) {
  matching_geoms <- 
    map(geom_aes_defaults(), name) %>%
      compact() %>%
      keep(~ !is.na(.) & . == old_aes)
  geoms <- gsub("^Geom(.*)", "\\1", names(matching_geoms))
  walk(geoms, update_geom_defaults, setNames(list(new_aes), name))
}

Now you can replace colors systematically, e.g. turn black into red by

replace_geom_aes_defaults("colour", "black", "red")

or even replace fill colors (for bar plots) by

replace_geom_aes_defaults("fill", "grey35", "red")

Chrysa