How do I change the colour palette in the dabest_plot function in dabestr_2023.9.12

Question

Just wanting to add a custom palette to this visualization. I have tried adding it manually using palette = c("colour1", "colour2",...) and using custom_palette. Nothing is working. Is there a way to extract this data and run it in ggplot2?

Here is the code I am working with: #Multi-group analysis in dabestr

multi_groups <- load(data, x = Treatment, y = Total_CO2,
idx = list(
c("GB-L", "LDPE-L"),
c("GB-H", "LDPE-H"),
c("GF-L", "PA-L"),
c("GF-H", "PA-H") ))

print(multi_groups)

multi_groups.mean_diff <- mean_diff(multi_groups)
print(multi_groups.mean_diff)
multi_groups.mean_diff

#Plots

dabest_plot(multi_groups.mean_diff) <--- THIS IS WHERE I WANT TO CHANGE THE COLOUR PALETTE

enter image description here

dabest_plot(multi_groups.mean_diff, palette = c("colour1", "colour2"... etc) - did not change the colours 

Answer 1

If you look at the documentation for ?dabest_plot, you will see that it only describes two named arguments, neither of which control the color, but the function also allows extra arguments to be passed via ....

... Adjustment parameters to control and adjust the appearance of the plot. (list of all possible adjustment parameters can be found under plot_kwargs)

We therefore need to look up the documents for ?plot_kwargs to find out how to control the color palette. Here we find that there is an argument called custom_palette:

custom_palette Default "d3". String. The following palettes are available for use: npg, aaas, nejm, lancet, jama, jco, ucscgb, d3, locuszoom, igv, cosmic, uchicago, brewer, ordinal, viridis_d.

Essentially, these are the palettes you are stuck with. Although the final object produced by the dabest_plot function is a ggplot object, it is actually a cowplot type ggplot, meaning one cannot simply add new color and fill scales to it.

To be fair, there are several palettes to choose from, all of which give quite a professional look.

Here's an example:

library(dabestr)

dabest_obj <- load(non_proportional_data,
                   x = Group, y = Measurement,
                   idx = c("Control 1", "Test 1", "Test 2", "Test 3"))

dabest_obj.mean_diff <- mean_diff(dabest_obj)

dabest_plot(dabest_obj.mean_diff, TRUE)

And here with a different palette:

dabest_plot(dabest_obj.mean_diff, TRUE, custom_palette = "igv")

From looking at the code of the package, it doesn't look as though it would be terribly difficult to add in a fully customizable palette, but it seems that this is not a priority for the authors right now.