I’m making a ggplot2 graph in which I’ve mapped an extremely wide range of numbers to the color. The problem is that my numbers are log-normally distributed, so I have a bunch of numbers all getting set to the same color since the gradient in ggplot2 assumes that the values are on a pretty similar linear scale. Here are some example data:

DF <- data.frame(A = 1:10, 
             B = 1:10, 
             C = 10^seq(-5, 4))

When I use the regular ggplot2 settings, all but my very last point are dark blue.

ggplot(DF, aes(x = A, y = B, color = C)) +
   geom_point()

If I log transform these values, the colors look better, but now my legend shows log-transformed values rather than the actual values.

ggplot(DF, aes(x = A, y = B, color = log10(C))) +
   geom_point()

I also tried using scale_color_gradientn in hopes that maybe I could set the numerical breaks the way I wanted, but that really did not work. The extremely small values are now all grey or missing.

ggplot(DF, aes(x = A, y = B, color = C)) +
   geom_point() +
   scale_color_gradientn(colors = grDevices::blues9, 
                         values = 10^seq(-5, 4))

Is there a way to make the colors of my gradient be on a log10 scale but the labels still be the actual values?