R -ggplot-在图形上绘制p值

Question

I have a question concerning the plotting of p-values in a violin plot in using ggplot in R. I have a dataframe which contain value ordered by groups : 1000 / 2000 / 3000/ .../n

I plot a violin plot from the dataframes (see example below).

My problem is that the last value is equal to length of dataframe. In some case can be 14470 in another dataframe it can be 16043 or 13789.

I want to plot p-value (wilcoxon test) on my plot by comparing violin 2 by 2.

What I did :

my_comparisons_1000 <- list \
(c("1000", "2000"),c("2000", "3000"),\
c("3000", "4000"),c("4000","5000"),\
c("5000","6000"),c("6000","7000"),\
c("7000","8000"),c("8000","9000"),\
c("9000","10000"),c("10000","11000"),\
c("11000","12000"),c("12000","13000"),\
c("13000","14000"))

fig_1000<-ggplot(violin, aes(x=range_1000, y=mean_region))+
    geom_violin(scale = "width",adjust = .5,fill='#A4A4A4', color="darkred")+
    geom_boxplot(width=0.1,outlier.shape = NA) + theme_minimal()+
    scale_x_discrete(labels=c(seq(1000,length(violin[,1]),by=1000), length(violin[,1])))+
    stat_summary(fun.y=mean, geom="point",size=1,color="red",aes(shape="Mean")) +

    stat_compare_means(comparisons = my_comparisons_1000,label.y = 14)+ # Add pairwise comparisons p-value

    theme(axis.text.x = element_text(angle = 90, hjust = 1))+
    guides(colour=guide_legend(order=1), shape=guide_legend(title=NULL, order=2)))

Goal

What I want is to do something shorter than my_comparisons_1000 and which fit the length of my dataframe different dataframes.

In this exemple I have groups of 1000 but I also have dataframes with groups of 500.

Actually I just have to improve 'my_comparisons_1000'

is there a way to generate several vector by step (1000) ? Something like rep or seq but I can't find it.

Answer 1

Something like this?

library(tidyverse)
library(ggpubr)
tidyiris <- iris %>% gather(key, value, -Species)
num_pairs <- length(unique(tidyiris$key)) - 1
my_comparisons <- map(seq(1, num_pairs, 1), ~c(.x, .x+1))
ggplot(tidyiris, aes(key, value)) + geom_violin() + 
  stat_compare_means(comparisons = my_comparisons)

For your data, it would be:

my_comparisons <- map(seq(1000, violin$range_1000 - 1000, 1000), ~c(.x, .x + 1000))