Support statistical annotation for grouped plots

[LeiGuo0812]

Hi,
I sincerely appreciate the tidyplots package. It's a fantastic tool for creating beautiful and tidy plots efficiently.

I particularly like the stat_pvalue_manual function in the ggpubr package, which is great for handling complex grouped or faceted plots. However, it seems that the current add_test_pvalue function in tidyplots only supports statistical tests for a single factor.

I tried using the add function to directly apply the stat_pvalue_manual, but encountered some errors. Is it possible to enhance tidyplots to support more complex grouped or faceted plots? Or is there currently a workaround for these scenarios?

Below are the test codes I used:

library(ggpubr)
library(rstatix)
library(tidyplots)

df <- ToothGrowth
df$dose <- as.factor(df$dose)
head(df, 3)

stat.test <- stat.test <- df %>%
  group_by(dose) %>%
  t_test(len ~ supp) %>%
  add_xy_position(x = "dose", dodge = 0.8)
  
stat.test2 <- df %>%
  t_test(len ~ dose) %>% 
  add_xy_position(x = "dose")

stat.test3 <- df %>%
  group_by(supp) |> 
  t_test(len ~ dose) %>% 
  add_xy_position(x = "dose", group = 'supp')

# it worked fine for ggplot
ggplot(df, aes(x = dose, y = len, color = supp)) +
  geom_boxplot() +
  stat_pvalue_manual(
    stat.test,  label = "p", tip.length = 0
  ) +
  stat_pvalue_manual(
    stat.test2,  label = "p", tip.length = 0.02,
    step.increase = 0.05
  ) +
  stat_pvalue_manual(
    stat.test3,  label = "p", tip.length = 0.02,
    step.increase = 0.05, step.group.by = 'supp'
  )

# but get error when using tidyplots
tidyplot(df, x = dose, y = len, color = supp)  |> 
  add_boxplot() |> 
  add(add_pvalue(
    stat.test,  label = "p", tip.length = 0
  ))  |>
  add(stat_pvalue_manual(
    stat.test2,  label = "p", tip.length = 0.02,
    step.increase = 0.05
  ))  |>
  add(stat_pvalue_manual(
    stat.test3,  label = "p", tip.length = 0.02,
    step.increase = 0.05, step.group.by = 'supp'
  ))

#Error:
#! Problem while computing aesthetics.
#ℹ Error occurred in the 2nd layer.
#Caused by error:
#! can not find object 'supp'

Thank you so much for your hard work on this amazing package, and I really look forward to your response!

[jbengler]

Hi @LeiGuo0812

Thank you for using tidyplots!

Your error occurs because ggpubr::stat_pvalue_manual() fails when both color and fill aesthetics are provided.
I can replicate the issue in your ggplot2 code by adding fill = supp and fix it in tidyplots by setting fill = NA. See the example below.

A bit of background: tidyplots automatically assigns the same variable to both color and fill unless explicitly specified otherwise.

Best
Jan

# adding aes(fill = supp) breaks the ggplot
# does not work
ggplot(df, aes(x = dose, y = len, color = supp, fill = supp)) +
  geom_boxplot() +
  stat_pvalue_manual(
    stat.test,  label = "p", tip.length = 0
  ) +
  stat_pvalue_manual(
    stat.test2,  label = "p", tip.length = 0.02,
    step.increase = 0.05
  ) +
  stat_pvalue_manual(
    stat.test3,  label = "p", tip.length = 0.02,
    step.increase = 0.05, step.group.by = 'supp'
  )

# adding aes(fill = NA) rescues the tidyplot
# just works
tidyplot(df, x = dose, y = len, color = supp, fill = NA) %>% 
  add_boxplot() %>% 
  add(stat_pvalue_manual(
    stat.test,  label = "p", tip.length = 0
  )) %>%
  add(stat_pvalue_manual(
    stat.test2,  label = "p", tip.length = 0.02,
    step.increase = 0.05
  )) %>% 
  add(stat_pvalue_manual(
    stat.test3,  label = "p", tip.length = 0.02,
    step.increase = 0.05, step.group.by = 'supp'
  ))

[LeiGuo0812]

Dear @jbengler
Thank you very much for your quick response! It perfectly solved my problem.
I have another question: if I need to specify both color and fill mapping in ggplot2, I can avoid the error by moving the mapping down to the sub-layers. I wonder if there is a similar way to handle this in tidyplots?

[jbengler]

Unfortunately not. In an attempt to keep things simple, the mappings in tidyplots are restricted to the tidyplot() call.

Since the error also occurs in plain ggplot2 + ggpubr, one could report this issue in ggpubr. Maybe they are able to fix it for all ggpubr users.

[jbengler]

I opened this ggpubr issue kassambara/ggpubr#621