How to Create a Beautiful Plots in R with Summary Statistics Labels

Prerequisites

Load required R packages

library(tidyverse)
library(rstatix)
library(ggpubr)

Data preparation:

# Demo data
data("ToothGrowth")
df <- ToothGrowth
df$dose <- as.factor(df$dose)
# Add random QC column
set.seed(123)
qc <- rep(c("pass", "fail"), 30)
df$qc <- as.factor(sample(qc, 60))
# Inspect the data
head(df)

##    len supp dose   qc
## 1  4.2   VC  0.5 fail
## 2 11.5   VC  0.5 pass
## 3  7.3   VC  0.5 fail
## 4  5.8   VC  0.5 pass
## 5  6.4   VC  0.5 pass
## 6 10.0   VC  0.5 pass

Basic box plots with add summary statistics

In the following R code, possible values for the argument ggfunc are the ggpubr R package functions, including: ggboxplot, ggviolin, ggdotplot, ggbarplot, ggline, etc. It can be also any other ggplot function that accepts the following arguments: data, x, color, fill, palette, ggtheme, facet.by.

# Basic plot
ggsummarystats(
  df, x = "dose", y = "len", 
  ggfunc = ggboxplot, add = "jitter"
  )

# Color by groups
ggsummarystats(
  df, x = "dose", y = "len", 
  ggfunc = ggboxplot, add = "jitter",
  color = "dose", palette = "npg"
  )

# Compute summary statistics
summary.stats <- df %>%
  group_by(dose) %>%
  get_summary_stats() %>%
  select(dose, n, median, iqr)
summary.stats

# Create a boxplot
bxp <- ggboxplot(
  df, x = "dose", y = "len", add = "jitter", 
  ggtheme = theme_bw()
)

# Visualize the summary statistics
summary.plot <- ggsummarytable(
  summary.stats, x = "dose", y = c("n", "median", "iqr"),
  ggtheme = theme_bw()
  ) +
  clean_table_theme()

# Combine the boxplot and the summary statistics plot
ggarrange(
  bxp, summary.plot, ncol = 1, align = "v",
  heights = c(0.80, 0.20)
  )

Grouped plots with summary table

# Grouped plots
ggsummarystats(
  df, x = "dose", y = "len", 
  ggfunc = ggboxplot, add = "jitter",
  color = "supp", palette = "npg"
  )

# Change plot type to violin
ggsummarystats(
  df, x = "dose", y = "len", 
  ggfunc = ggviolin, add = c("jitter", "median_iqr"),
  color = "supp", palette = "npg"
  )

# Create barplot
ggsummarystats(
  df, x = "dose", y = "len", 
  ggfunc = ggbarplot, add = c("jitter", "median_iqr"), position = position_dodge(),
  color = "supp", palette = "npg"
  )

# Create line plots
ggsummarystats(
  df, x = "dose", y = "len", 
  ggfunc = ggline, add = c("jitter", "median_iqr"), 
  color = "supp", palette = "npg"
  )

ggsummarystats(
  df, x = "supp", y = "len", 
  ggfunc = ggboxplot, add = c("jitter"), 
  color = "dose", palette = "npg"
  )

Multipanel plots with summary table

Key arguments:

• facet.by: character vector, of length 1 or 2, specifying grouping variables for faceting the plot into multiple panels. Should be in the data.
• labeller: Character vector. Possible values are one of label_both (panel labelled by both grouping variable names and levels) and label_value (panel labelled with only grouping levels).

Create panels according to one grouping variable:

ggsummarystats(
  df, x = "dose", y = "len", 
  ggfunc = ggboxplot, add = c("jitter"), 
  color = "dose", palette = "jco",
  facet.by = "supp", labeller = "label_value",
  ggtheme = theme_bw(), legend = "top"
)

Create panels according to two grouping variables

ggsummarystats(
  df, x = "dose", y = "len", 
  ggfunc = ggboxplot, add = c("jitter"), 
  color = "dose", palette = "jco",
  facet.by = c("supp", "qc"), labeller = "label_both", 
  ggtheme = theme_bw(), legend = "top"
)

Create independent panels using the argument free:

ggsummarystats(
  df, x = "dose", y = "len", 
  ggfunc = ggboxplot, add = c("jitter"), 
  color = "dose", palette = "jco",
  facet.by = c("supp", "qc"), labeller = "label_both", 
  free.panels = TRUE,
  ggtheme = theme_bw(), legend = "top"
)

Build step by step a custom multipanel plot

Create a multipanel box plot using one grouping variable (supp):

# Group the data by supp
# Apply the function ggsummarystats to each subset
df.grouped <- df %>%
  df_split_by(supp, label_col = "panel", labeller = df_label_both) %>%
  mutate(plot_list = map(
    data, ggsummarystats, x = "dose", y = "len",
    ggfunc = ggbarplot, 
    add = c("jitter", "median_iqr"), 
    facet.by = "panel"
    )
  )
df.grouped

## # A tibble: 2 x 4
##   supp  data              panel   plot_list 
##   <fct> <list>            <fct>   <list>    
## 1 VC    <tibble [30 × 4]> supp:OJ <ggsmmrys>
## 2 OJ    <tibble [30 × 4]> supp:VC <ggsmmrys>

# Print the plots
plot_list <- df.grouped$plot_list
class(plot_list) <- c("ggsummarystats_list", "list")
print(plot_list)