R | ggplot2 | bar in barplot does not start at the right value

Question

I want to visualize the reference ranges of several liver enzymes (for example GOT and GPT) that I calculated with two programs "kosmic" and "RLE" using ggplot2.

I do not understand why the bars always start at 0, even if the lower range is for example 16.02.

How do I need to change my code so the minimum and maximum values of the bars look like that:

[16.02,45.46] [9.16,60.52] [16.10,68.90] and [9.30,64.40].

Thank you in advance!

 #install.packages("ggplot2") library(ggplot2) program <- c(rep("kosmic",4),rep("RLE",4)) value <- c(16.02,45.46,9.16,60.52,16.1,48.9,9.3,64.4) parameter <- c(rep("GOT",2),rep("GPT",2),rep("GOT",2),rep("GPT",2)) table1 <- data.frame(program,value,parameter) p <- ggplot(table1, aes(parameter,value, fill = program))+ geom_bar(position="dodge", stat="identity") print(p)

I am looking for something like this:

Answer 1

Are you looking for something like this?

library(dplyr)
table1 %>%
  group_by(parameter, program) %>%
  summarize(min = min(value), 
            median = median(value),
            max = max(value), .groups = "drop") %>%
ggplot(aes(interaction(parameter,program), fill = program))+
  geom_tile(aes(y = median, height = max-min), width = 0.6)

Edit: Okay this is hacky, but:

table1 %>%

 # example of reordering the parameters
 mutate(parameter = fct_relevel(parameter, "GPT", after = 0)) %>%
  # forcats offers a variety of fct_*** functions to change factors
  # (factors are a data type that can separately store labels and ordering)

  group_by(parameter, program) %>%
  summarize(min = min(value), 
            median = median(value),
            mean = mean(value),
            max = max(value), .groups = "drop") %>%
  ggplot(aes(parameter, mean, color = program))+
  geom_errorbar(aes(ymin = min, ymax = max), 
                position = position_dodge(width = 0.3), size = 10,
                width = 0) + 

  # control the legend so the key squares aren't gigantic to match the error bar widths
  guides(colour = guide_legend(override.aes = list(size=8))) +

  # example of assigning different colors. 
  # a variety of scale_color_* functions are available
  scale_color_manual(values = c("kosmic" = "#cc5588", "RLE" = "#779988"))

A downside of this is that the width/spacing of the bars will vary depending on your graphic output aspect ratio, so to use it might take some fiddling to get as you want.

Answer 2

Based on what you want, I'd suggest a box plot instead of a bar plot:

ggplot(table1, aes(x = parameter, y = value, fill = program, color = program)) +
    geom_point(position = position_jitterdodge()) + 
    geom_boxplot(outlier.shape = NA, color = 'black')