Selecting hue/brewer colours manually in ggplot2

Question

I am trying to produce progressive plots using the ggplot2 library in R.

What I mean by 'progressive' is that, for presentation purposes, I want to add lines to a plot one at a time, so I generate a single plot with many lines multiple times, each with an extra plotted line.

In ggplot2, using scale_colour_hue() for example works well, except that the colours for each line change from plot to plot. I want to keep the colour constant (i.e. plot 1 line 1 has colour X and in plot 5, line 1 still has colour X...etc). I can achieve that by manually declaring colours using scale_colour_manual() but I feel like I'm limited by the aesthetics of these colours ("red", "blue", etc). Do I really have to find the hex value for each of those colours in say the hue palette or is there an easier way out, where I can instruct scale_colour_hue() function to use its palette in a fixed order everytime it adds a new line?

Thanks, appreciate the help!

Answer 1

If you know beforehand the total number of lines you wish to add, you can create a scale with the appropriate colours and names for use with scale_colour_manual

library(RColorBrewer)
library(plyr)
# a function to create a brewer palette of any length (kittens will be killed)
myBrewerPal <- function(pal = 'Spectral'){
  colorRampPalette(brewer.pal(name = pal, n = 11))
}

# function to create a named vector of colours for a given 
# vector x
create.palette <- function(x, pal = 'Spectral'){
  ux <- sort(unique(x))
  n <- length(ux)
  setNames(myBrewerPal(pal)(n), ux)

}

# create a manual scale with the named values part
myPal <- scale_colour_manual(name = 'Gear', values = create.palette(factor(mtcars$gear)))


# the base plot (no lines)
baseplot <- ggplot(mtcars, aes(x=hp, y = disp, colour = factor(gear))) + myPal
# 1 line
baseplot + geom_line(subset = .(gear==3))

# 2 lines (gear = 3 is consistent)
baseplot + geom_line(subset = .(gear %in% 3:4))