merging two different colors with ggplot2 geom_point plot

[Eric Foss]

(I also posted this on the R forum at talkstats.com.)

In cell biology, people often publish images in which one protein is labeled green and a second protein is labeled red, and then they show an image of the green protein only, then the red protein only, and then a "merge" to show where both proteins are present.  The merge ends up showing mostly green where only protein 1 is, mostly red where only protein 2 is, and yellow where they both are, and you can see gradients where there are increasing levels of protein 2 in a background of constant protein 1, etc.  I want to do the same sort of thing with ggplot, with the x and y axis showing locations of two measurements and the intensity of color indicating the magnitude of the measurement.  In the attached example, measurements 1 and 2 are similar over most of x and y coordinates, but measurement 1 is a bit higher from y axis 50 to 100 and measurement 2 is a bit higher from y axis 75 to 125.  In my example, graph1 gives me what I want (albeit in blue rather than red), but then I show an unsuccessful "graph2" in which measurements 2 simply overwrite measurements 1. 

 

 I hope that with the attached script, this confusing question becomes more clear.  Can anyone tell me how to make the desired merge of two different colors (ideally red for 1 and green for the other)?  (Any constructive criticism of the somewhat ugly way in which I've created the data frame would also be welcome.)  

 

Thanks. 

 

Eric

 

[Eric Foss]

Here is a little more discussion of my question at stackoverflow

[David Kahle]

Maybe something like this?

library("tidyverse"); theme_set(theme_bw())

set.seed(8)

# make the dataset

n_x <- 100

n_y <- 250

list(x = 1:n_x, y = 1:n_y) %>% 

  cross_df() %>% 

  mutate(

    z1 = rnorm(n_x*n_y) + 1.5*(50 < y & y <= 100),

    z1 = as.vector(scale(z1)),

    z2 = rnorm(n_x*n_y) + 1.5*(75 < y & y <= 125),

    z2 = as.vector(scale(z2))

  ) -> df

  

# plot both datasets (z1 and z2)

df %>% 

  gather(which_z, value, -x, -y) %>% 

  ggplot(aes(x, y, fill = value)) +

    geom_raster() +

    facet_wrap(~ which_z)

# look only at z1

df %>% 

  ggplot(aes(x, y, fill = z1)) +

  geom_raster() +

  scale_fill_gradient2(low = "red", mid = "white", high = "green")

df %>% 

  ggplot(aes(x, y, fill = z1, alpha = (z1 >= 0))) +

  geom_raster() +

  scale_fill_gradient2(low = "red", mid = "white", high = "green") +

  scale_alpha_manual(

    values = c("TRUE" = 1, "FALSE" = 0),

    guide = FALSE

  )

df %>% 

  ggplot(aes(x, y, fill = z1, alpha = (z1 <= 0))) +

  geom_raster() +

  scale_fill_gradient2(low = "red", mid = "white", high = "green") +

  scale_alpha_manual(

    values = c("TRUE" = 1, "FALSE" = 0),

    guide = FALSE

  )

Cheers

david.

===================================

David J. Kahle, Ph.D.
Associate Professor
Department of Statistical Science
Baylor University
www.kahle.io

--
--
You received this message because you are subscribed to the ggplot2 mailing list.
Please provide a reproducible example: https://github.com/hadley/devtools/wiki/Reproducibility
 
To post: email ggp...@googlegroups.com
To unsubscribe: email ggplot2+u...@googlegroups.com
More options: http://groups.google.com/group/ggplot2

---
You received this message because you are subscribed to the Google Groups "ggplot2" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ggplot2+u...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.