putting significance levels over various facets-CodePudding

The

My question is: How do I create a significance level over two values that stem from two different facets? For instance, '1' in facet 1 and '1' in facet 2 in the example. I do not manage to set the xmax values to grap the subsequent facet as the x-axis is repeated in all facets. The desired outcome would draw one single line from A=1 to A=1 from one facet to the other and place the sign * in the middle of the line. Similar to the second plot I posted but at the positions of the first plot.

Switching the fill and facet variables is not an option.

As you may see, I have already tried to ungroup the corresponding variable C with data = group_by(...) but this did not help.

CodePudding user response：

According to this issue there seems no inbuilt option. You could use ggplot_build to modify the data behind your first plot, so that you can connect the line from panel 1 to panel 2. For this you need to change the x and xend values of the plot and set the layout clip to 'off' to make lines across facets. Here is a reproducible example:

library(ggplot2) library(dplyr) library(ggpubr) set.seed(7) # reproducibility value <- sample(1 :50,100,replace = T) B <- as.factor(rep(c(1,2,2,1), 50) ) C <- as.factor(rep(c(1,2),each=100) ) A <- as.factor(rep(c(4,4,3,3,2,2,1,1), 25) ) dt1 <- data.frame(A,B,C,value) %>% group_by(A,B,C) p <- ggplot(dt1, aes(x=A, y=value, fill=(B))) theme_classic() facet_wrap(vars(C)) geom_boxplot(position='dodge',outlier.size=0) geom_point(size=0.5,position=position_jitterdodge(0.2)) geom_signif(data = dt1 %>% group_by(A,B), y_position = 55, xmin = 0.8, xmax = 4.8, annotation = rep("*"), tip_length = 0.02) # Use ggplot_build q <- ggplot_build(p) # Modify data q$data[[3]]$xend <- c(0.8, 5.8, 5.8, 0.8, 0.8, 0.8) q$data[[3]]$x <- c(0.8, 0.8, 6.8, 0.8, 0, 0.8) q$data[[3]]$annotation <- c(rep('*', 3), rep('', 3)) # Add data back to plot q <- ggplot_gtable(q) # Change clip layout q$layout$clip <- 'off' # plot data again plot(q)

^{Created on 2022-10-02 with reprex v2.0.2}

As you can see, there is a comparison across facets.