Selecting divergent colors

It will be hard to come up with 20 clearly distinguishable colors.  Check
out the website http://colorbrewer2.org/ and the R package RColorBrewer.
It does not have a 20-color palette, but it does have some 8- to 12-color
palettes that are very nice.

library(RColorBrewer)
display.brewer.all(n=NULL, type="all", select=NULL, exact.n=TRUE)

You could use these colors in combination with line type to build up to 72
unique combinations.  For example ...

nuniq <- ncol(mat)
mycols <- rep(brewer.pal(12, "Set3"), length=nuniq)
myltys <- rep(1:6, rep(12, 6))[1:nuniq]

for(k in 1:nuniq){
plot(density(mat[,k]), col=mycols[k], xlab='', ylab='', axes=F, main=F,
lwd=3, lty=myltys[k])
par(new=TRUE)
}
legend('topright', legend=snames, col=mycols, lty=myltys, lwd=3)

Jean



On Mon, Jun 10, 2013 at 12:33 PM, Brian Smith <bsmith030465 at gmail.com>wrote:

Hi,

I was trying to make a density plot with 13 samples. To distinguish each
sample, it would be good if each color is as different as possible from the
other colors. I could use the built in function, but that does not do more
than 8 colors and then goes back to recycling the cols. If I use a palette,
then it is really difficult to distinguish between the colors.

So, is there a way that I can select a large number of colors (i.e. perhaps
20) that are as different from each other as possible?

Here is my example code using the palette:

**********************
mat <- matrix(sample(1:1000,1000,replace=T),nrow=20,ncol=20)
snames <- paste('Sample_',1:ncol(mat),sep='')
colnames(mat) <- snames

mycols <- palette(rainbow(ncol(mat)))

for(k in 1:ncol(mat)){
 plot(density(mat[,k]),col=mycols[k],xlab='',ylab='',axes=F,main=F)
 par(new=T)
}

legend(x='topright',legend=snames,fill=mycols)

****************************

thanks!

       [[alternative HTML version deleted]]

	[[alternative HTML version deleted]]