How to reorder only specific names on x-axis of boxplot in R?

Question

I have a dataframe test like below:

dput(test)

structure(list(Groups = c("Group1", "Group2", "Group3", "Group4", 
"Group5", "Group6", "Group7", "Group8", "Group9", "Group10", 
"Group11", "Group12", "Group13", "Group14", "Group15", "Group16", 
"Group17", "Group18", "Group19", "Group20", "Group21", "Group22"
), Disease = c("Brain", "Brain", "Brain", "Blood", "Blood", "Esophagus", 
"Esophagus", "Esophagus", "Brain", "Brain", "OE", "control", 
"OE", "control", "OE", "control", "PE", "PE_X", "PE_X", "PE", 
"PE_X", "PE"), variable = c("Name", "Name", "Name", "Name", "Name", 
"Name", "Name", "Name", "Name", "Name", "Name", "Name", "Name", 
"Name", "Name", "Name", "Name", "Name", "Name", "Name", "Name", 
"Name"), value = c(1.079825876, 0.236961206, 0.286498286, 0.374978442, 
3.620160544, 1.876875376, 0.293402656, 0.176208121, 0.622282653, 
1.373705338, 9.235592994, 1.437889832, 8.70900915, 1.772903362, 
9.070885831, 1.792899823, 10.29580836, 1.373281466, 4.210242765, 
0, 7.331498976, 14.11415563)), class = "data.frame", row.names = c(NA, 
-22L))

Using below code I made a boxplot:

ggplot(data= subset(test, variable == "Name")) + 
  geom_boxplot(aes(x=Disease, y=value, fill=Disease), outlier.shape=NA) +
  geom_jitter(aes(x=Disease, y=value, fill=Disease), position=position_dodge(0.2)) +
  theme_classic(base_size = 12) + xlab("") + ylab("value")

Using the below code I reordered the x-axis names (Disease) based on the column value.

ggplot(data= test) + 
  geom_boxplot(aes(x=reorder(Disease,value, na.rm=TRUE), y=value, fill=Disease), outlier.shape=NA) +
  geom_jitter(aes(x=Disease, y=value, fill=Disease), position=position_dodge(0.2)) +
  theme_classic(base_size = 12) + xlab("") + ylab("value")

Question: I would like to reorder only Blood, Brain, Eophagus Diseases on x-axis and keep the rest of the Diseases on the right side of the plot. How to do this?

You can always transform the Disease variable to factor and change the order manually by writing the levels desired_order <- c(the vector of the order you want) data$category <- factor(test$Disease, levels = desired_order) — Lucca Nielsen, Jul 20 '23 at 19:28
I don't want to order all names of x-axis, want to order only few names based on the column value — beginner, Jul 20 '23 at 20:23
Could you please tell what the exact order should be of names on your x-axis? — Quinten, Jul 21 '23 at 07:58

TarJae · Answer 1 · 2023-07-20T20:27:22.860

3

Update after clarification:

library(ggplot2)
library(dplyr)

# medians
medians <- test %>% 
  filter(Disease %in% c("Esophagus", "Brain", "Blood")) %>%
  group_by(Disease) %>%
  summarize(median = median(value, na.rm = TRUE)) %>%
  arrange(median) %>%
  pull(Disease)

# rest of the levels
rest <- setdiff(unique(test$Disease), medians)

# Combine and reorder
new_order <- c(medians, rest)

# transform disease column to factor with the new levels
test$Disease <- factor(test$Disease, levels = new_order)

ggplot(data = subset(test, variable == "Name")) + 
  geom_boxplot(aes(x = Disease, y = value, fill = Disease), outlier.shape = NA) +
  geom_jitter(aes(x = Disease, y = value, fill = Disease), position = position_dodge(0.2)) +
  theme_classic(base_size = 12) +
  xlab("") +
  ylab("value")

First answer: Here is a solution using fct_relevel from forcats package:

library(forcats)
library(dplyr)
library(ggplot2)

# your order
first_levels <- c("Blood", "Brain", "Esophagus")

# adjusting levels 
test %>% 
  mutate(Disease = fct_relevel(Disease, first_levels)) %>% 
  filter(variable == "Name") %>%
  ggplot() +
  geom_boxplot(aes(x = Disease, y = value, fill = Disease), outlier.shape = NA) +
  geom_jitter(aes(x = Disease, y = value, fill = Disease), position = position_dodge(0.2)) +
  theme_classic(base_size = 12) + 
  xlab("") + 
  ylab("value")

edited Jul 20 '23 at 20:27

answered Jul 20 '23 at 19:35

TarJae

72,363
6
19
66

1

sorry, but I want to reorder Blood, Brain and Esophagus. In the above plot it doesn't look like they are ordered based on the value. – beginner Jul 20 '23 at 20:11
Please see my update! – TarJae Jul 20 '23 at 20:27
but what if I have multiple variables? the order might change right. Could you please tell what to do if I have multiple variables. – beginner Jul 20 '23 at 20:35
The order of the variables to be changed is defined here `medians <- test %>% filter(Disease %in% c("Esophagus", "Brain", "Blood")) %>%` by getting the medians of these defined variables we will order them by their medians, the rest will keep as it is ! – TarJae Jul 20 '23 at 20:37
no, what I mean is you can see column `variable` I have multiple names in that column, not just `Name`. There are many like `Name2`, `Name3`,`Name4` etc... until 100 – beginner Jul 20 '23 at 20:39
did you get my point? – beginner Jul 20 '23 at 20:52
no sorry. maybe you could explain? – TarJae Jul 20 '23 at 21:57

How to reorder only specific names on x-axis of boxplot in R?

1 Answers1