How to add a color frame to a plot?

Question

For some reasons I want to add a frame to a plot like this:

Can I do this in plot? ggplot or qplot solutions are also welcome. Thank you.

Answer 1

set.seed(123); 
plot(x=rnorm(100, sd=1000), y=rnorm(100, sd=1000) ,ylab="", xlab="")
rect(xleft=par("usr")[1]*1.25, ybottom=par("usr")[3]*1.4, 
     xright=par("usr")[2]*1.1,ytop=par("usr")[4]*1.2, 
     lwd=5, border="orange", xpd=TRUE)

Answer 2

qplot(x= disp , y = wt , data = mtcars) +
  theme(plot.background = element_rect(colour="#CF8057",size=10))

Answer 3

You can draw a solid rectangle, then using print(.., vp=..) plot your ggplot over it, shrinking it down slightly.

Here is an example in a nice little function:

borderize <- function(plotObj, thick=2, color="orange", alpha=0.8) {
  # thick should be a value between (1, 100)
  # alpha should be a value between (0, 1)

  # these could be modified for separate width/height thicknesses
  wd <- ht <-  (100 - thick) / 100
  x <- (1 - wd)  / 2 
  y <- (1 - ht) / 2

  # create a solid rectangle.  The plot will go over it. 
  grid.rect(y = 1, height = 1, just = c("center", "top"), gp=gpar(fill=color, alpha=alpha, lwd=0))

  # create the viewport
  vp.inner <- viewport(height=unit(ht, "npc"), width=unit(wd, "npc"), just=c("left","bottom"), y=y, x=x)

  print(plotObj, vp=vp.inner)

}

eg:

myPlot <- ggplot(iris,aes(Sepal.Length,Sepal.Width)) + 
                 geom_point(aes(color=Species,shape=Species))

borderize(myPlot, thick=5, color="#CF8042")

Note that you can also modify plot.background and panel.background with theme() in ggplot2.
However, this will impact your labels and legend, depending on the thickness of border, font size etc. Thats why I prefer to use viewport.

eg:

plot.bg  <- theme(plot.background=element_rect(color="red", size=12))
panel.bg <- theme(panel.background=element_rect(color="blue", size=12))

plotObj + panel.bg + plot.bg

red border is plot, blue border is panel

NOTICE HOW THE BORDER EATS INTO THE LABELS

The advantage of using plot.background however is that you can save the whole graph as an object; something you cannot do using the borderize method above.