r - Convex hull on horizontal and vertical error bars

Question

I have a ggplot with means and both horizontal and vertical error bars, and I'd like to add a convex hull that encompasses all the error bars - like so:

I have tried with stat_chull from ggpubr but am not sure how to specify the aesthetics when there is a xmin, xmax, ymin and ymax for the error bars of each point. Below is how I coded the plot specifying the stat_chull aes as the X and Y means as an example - I know this is incorrect but I think I am on the right track?

ggplot()+
  geom_point(data = df, aes(MeanX,MeanY)) +
  geom_errorbar(data = df, 
                mapping = aes(x = MeanX,
                              ymin = MeanY - SdY, 
                              ymax = MeanY + SdY), 
                width = 0, inherit.aes = FALSE)+
  geom_errorbarh(data = df, 
                 mapping = aes(y = MeanY,
                               xmin = MeanX - SdX,
                               xmax = MeanX + SdX),
                 height = 0, inherit.aes = FALSE)+
  stat_chull(data = df, aes(MeanX,MeanY))+
  theme_classic()

This gives the following plot:

I have also tried geom_polygon and got garbage.

Here is a the data:

df<-structure(list(Source = structure(1:5, .Label = c("A", "B", "C", "D", 
                                                      "E"), class = "factor"), MeanX = c(-18.7066666666667, 
                                                                                                                                -15.8769230769231, -16.8620689655172, -15.72, -17.4333333333333
                                                      ), SdX = c(1.61072554509115, 0.409201849758959, 1.04811067886951, 
                                                                       0.74057035077327, 1.15902257671425), MeanY = c(9.93666666666667, 
                                                                                                                            14.3230769230769, 9.22758620689655, 11.1, 13.7333333333333), 
                   SdY = c(1.03005970142791, 0.539116085686704, 0.504990221704281, 
                                 0.757187779440037, 1.05039675043925)), row.names = c(NA, 
                                                                                      -5L), class = "data.frame")

Any help is greatly appreciated!

标签： rggplot2convex-hull

Answer 1

Does the following work for you?

library(dplyr)

df %>%
  mutate(ymin = MeanY - SdY,
         ymax = MeanY + SdY,
         xmin = MeanX - SdX,
         xmax = MeanX + SdX) %>%
  ggplot(aes(x = MeanX, y = MeanY))+
  geom_point() +
  geom_errorbar(aes(ymin = ymin, ymax = ymax),
                width = 0)+
  geom_errorbarh(aes(xmin = xmin, xmax = xmax),
                 height = 0)+
  stat_chull(data = . %>%
               summarise(x = c(MeanX, MeanX, MeanX, xmin, xmax),
                         y = c(MeanY, ymin, ymax, MeanY, MeanY)),
             aes(x = x, y = y),
             geom = "polygon", colour = "black", fill = NA)+
  theme_classic()