Control point border thickness in ggplot

Question

When using ggplot, I can set shape to 21-25 to get shapes that have independent setting for the internal (fill) and border (col) colors, like so:

df <- data.frame(id=runif(12), x=1:12, y=runif(12))
ggplot(df, aes(x=x, y=y)) + 
  geom_point(aes(fill=id, size=id), colour="black", shape=21)

However, I can't figure out how to control the thickness of the shape borders, either setting them absolutely or as an aesthetic mapping. I note that if I set an lwd value, it overrides the size aesthetic:

ggplot(df, aes(x=x, y=y)) + 
  geom_point(aes(fill=id, size=id), colour="black", shape=21, lwd=2)

How do I control the border thickness?

Krahling

Answer 1

Starting in version 2.0.0 of ggplot2, there is an argument to control point border thickness. From the NEWS.md file:

geom_point() gains a stroke aesthetic which controls the border width of shapes 21-25 (#1133, @SeySayux). size and stroke are additive so a point with size = 5 and stroke = 5 will have a diameter of 10mm. (#1142)

Thus, the correct solution to this is now:

df <- data.frame(id=runif(12), x=1:12, y=runif(12))
ggplot(df, aes(x=x, y=y)) + 
  geom_point(aes(fill=id, size=id), colour="black", shape=21, stroke = 2)

Krahling

Can you add the output? – Halfon

For future compatibility and clarity it would be nice to say "Starting in version x.xx" rather than "the latest version". – Hydrastinine

Is there a way of controlling the thickness of the stroke based on the variables? Something along the lines of "scale_stroke_manual"? – Emplace

As a response to my own comment above, we can use scale_discrete_manual. See link here: github.com/tidyverse/ggplot2/issues/3507#issuecomment-526212978 – Emplace

Any ideas why stroke=0 still draws it in PDF but not in PNG? – Trisyllable

Answer 2

It feels a bit hacky but you can add a "background" set of dots with the size set to the aesthetic mapping plus some small constant to enlarge the border of the dots. Play with the constant to get the desired border width.

You'll also have to disable the size legend to stop it displaying the legend on the graph...

ggplot(df, aes(x=x, y=y)) + 
  geom_point(aes(size=id+0.5), colour="black" , show_guide = FALSE )+
  scale_size( guide = "none" )+
  geom_point(aes(fill=id, size=id), colour="black", shape=21)

Foah

Maybe hackey, but Hadley endorsed this method on Twitter – Krahling

@NoamRoss do you have the link for that? – Ferriferous

Answer 3

Another solution is to create the plot in R and then export it to a .svg file. The linewidth can then be changed using vector graphics editing software (e.g., Inkscape). This method is particularly useful when data points overlap.

Decca