R. Fixing labels in pie chart

Question

I am using this script:

# my data:
df <- data.frame(Somatic = c("PIK3CA", "TP53", "GATA3", "KMT2C", "PTEN", "NCOR1", "ARID1A", "RUNX1", "NF1", "TBX3", "FOXA1", "ERBB2", "AKT1", "PREX2", "CBFB", "INPPL1", "NSD3", "ESR1", "AXIN2", "RAD51C"),
                 Freq_mut = c(32.6, 32.6, 11.9, 9.3, 5.4, 4.7, 3.8, 3.8, 3.6, 3.1, 2.9, 2.8, 2.5, 2.4, 2.3, 1, 1, 0.8, 0.6, 0.6))

library(randomcoloR)

# set of 20 random colors
colors <- distinctColorPalette(20)

ggplot(df, aes(x = "", y = Freq_mut, fill = Somatic)) +
  geom_bar(width = 1, stat = "identity", color = "black") +
  coord_polar("y", start = 0) +
  scale_fill_manual(values = colors) +
  ggtitle("A") +
  labs(fill = "Somatic Gene") +
  theme(panel.background = element_blank(), plot.background = element_blank(),
        axis.text = element_blank(),
        axis.ticks = element_blank()) +
geom_label_repel(data = df,
                   aes(y = Freq_mut, label = paste0(Somatic)),
                   size = 4.5, nudge_x = 1, show.legend = FALSE)

But as you can see in the image, the labels are not in the correct position and are missing the rest of the genes. How Could I fix it and remove the "x" to the left of the plot? Thanks!

Answer 1

In addition to @Allan Camerons perfect solution to implement the specified chart layout, I'd encourage to reconsider whether that layout indeed makes it easy for the viewer to grasp the information you want to convey.

An alternative display might help the viewer to associate gene(?) with mutation frequency and keep mental note of the top/bottom five "mutators" at one glance (if that's the intended message). By having the x-axis stretch from 0-100 % you can maintain if not improve the pie charts 'fraction-of-whole' aesthetic (angles being harder to compare than lengths).

Example (using your data df):

library(forcats) ## for convenient factor operations

df |>
    mutate(Somatic = fct_reorder(Somatic, Freq_mut, mean)) |>
    ggplot(aes(x = Freq_mut, y = Somatic)) +
    geom_bar(aes(x = 100), stat = "identity", fill = '#e0e0e0') +
    geom_bar(stat = "identity") +
    geom_text(aes(x = Freq_mut,
                  label = sprintf("%2.1f  ", round(Freq_mut, 1))
                  ), adj = 0, nudge_x = 1, size = 3) +
    facet_wrap(~ Freq_mut < median(Freq_mut), 
               scales = 'free_y', ncol = 2
               ) +
    scale_x_continuous(breaks = c(0, 100)) +
    labs(x = 'mutation frequency (%)', y = '') +
    theme(
        panel.background = element_blank(),
        strip.text.x = element_blank(),
        axis.text.y = element_text(hjust = 1, size = rel(1)),
        panel.grid = element_blank()
    )

result:

Answer 2

I think you may be best with direct labelling here. Remember if you have a stacked bar graph, the y values are stacked, so your text labels also need to be stacked using position_stack

library(randomcoloR)
library(geomtextpath)

# set of 20 random colors
colors <- distinctColorPalette(20)

ggplot(df, aes(x = 1, y = Freq_mut, fill = Somatic)) +
  geom_bar(width = 1, stat = "identity", color = "black") +
  coord_polar("y", start = 0) +
  scale_fill_manual(values = colors, guide = "none") +
  ggtitle("A") +
  labs(fill = "Somatic Gene", x = NULL) +
  theme(panel.background = element_blank(), plot.background = element_blank(),
        axis.text = element_blank(),
        axis.ticks = element_blank()) +
  geom_textpath(data = df, position = position_stack(vjust = 0.5),
                   aes(x = 1.7, y = Freq_mut, label = paste0(Somatic)),
                   size = 4.5, show.legend = FALSE)

or

ggplot(df, aes(x = 1, y = Freq_mut, fill = Somatic)) +
  geom_bar(width = 1, stat = "identity", color = "black") +
  coord_polar("y", start = 0) +
  scale_fill_manual(values = colors, guide = "none") +
  ggtitle("A") +
  labs(fill = "Somatic Gene", x = NULL) +
  theme(panel.background = element_blank(), plot.background = element_blank(),
        axis.text = element_blank(),
        axis.ticks = element_blank()) +
  geom_labelpath(data = df, position = position_stack(vjust = 0.5),
                   aes(x = 1.7, y = Freq_mut, label = paste0(Somatic)),
                   size = 4.5, show.legend = FALSE) 

To add percentages, you can try:

ggplot(df, aes(x = 1, y = Freq_mut, fill = Somatic)) +
  geom_bar(width = 1, stat = "identity", color = "black") +
  coord_polar("y", start = 0, clip = "off") +
  scale_fill_manual(values = colors, guide = "none") +
  xlim(0.5, 2) +
  ggtitle("A") +
  labs(fill = "Somatic Gene", x = NULL) +
  theme(panel.background = element_blank(), plot.background = element_blank(),
        axis.text = element_blank(),
        axis.ticks = element_blank()) +
  geom_textpath(data = df %>% 
                  mutate(perc = scales::percent(Freq_mut/ sum(Freq_mut))), 
                position = position_stack(vjust = 0.5), hjust = 1,
                aes(x = 1.55, y = Freq_mut, 
                    label = paste0(Somatic, " (", perc, ")")),
                size = 4.5, show.legend = FALSE)

Answer 3

You should precalculate the positions of the labels so you can position the labels around the pie chart. Here is an option using geom_label_repel:

library(ggplot2)
library(dplyr)
library(ggrepel)
library(randomcoloR)

# set of 20 random colors
colors <- distinctColorPalette(20)
df %>%
  arrange(desc(Somatic)) %>%
  mutate(pos_freq_mut = cumsum(Freq_mut) - Freq_mut/2) %>%
  ggplot(aes(x = "", y = Freq_mut, fill = Somatic)) +
  geom_col(color = "black") +
  coord_polar("y") +
  geom_label_repel(aes(y = pos_freq_mut, label = paste0(Somatic)),
                   size = 4.5, nudge_x = 1, nudge_y = 1, show.legend = FALSE) +
  scale_fill_manual(values = colors) +
  ggtitle("A") +
  labs(fill = "Somatic Gene", x = "") +
  theme(panel.background = element_blank(), plot.background = element_blank(),
        axis.text = element_blank(),
        axis.ticks = element_blank()) 

^{Created on 2022-12-27 with reprex v2.0.2}