Making figures ready for publication with ggplot2

This tutorial offers a step-by-step guide for how to create publication-ready figures using ggplot2 and the data from palmerpenguins.

Install the package & data

# Install the package
remotes::install_github("allisonhorst/palmerpenguins")

# Load the package
library(palmerpenguins)

# Load the data into the Global Environment
data("penguins")

Meet the Penguins

Artwork by @allison_horst

What are culmen length & depth?

The culmen is the upper ridge of a bird’s bill. In the simplified penguins data, culmen length and depth are renamed as variables bill_length_mm and bill_depth_mm to be more intuitive. For this penguin data, the culmen (bill) length and depth are measured as shown below:

Artwork by @allison_horst

The steps for creating a beautiful scatter plot in ggplot2

First we will create a basic scatterplot of body_mass_g against bill_length_mm.

# Load the package
library(ggplot2)

ggplot(penguins, aes(body_mass_g, bill_length_mm))+ # this is the data
     geom_point() # here we add the points

Change the size of points

We can manually change the size of our datapoints. The points in the standard plot are quite small, so lets increase the size of the points with size = 3.

ggplot(penguins, aes(body_mass_g, bill_length_mm))+ 
     geom_point(size = 3) 

Change the shape of points

In ggplot2, it is possible to change the shape of the points. Here is a quick reference guide:

The shape of all datapoints can be changed with e.g. shape = 8.

ggplot(penguins, aes(body_mass_g, bill_length_mm))+
     geom_point(size = 3, shape = 8) 

Alternatively, we can change the shape of our points based on species with aes(shape = species).

ggplot(penguins, aes(body_mass_g, bill_length_mm))+ 
     geom_point(aes(shape = species), size = 3) 

Change the opacity of points

You can also change the opacity of the data points using alpha. Alpha values are required to be between 0 - 1 where 0 is transparent and 1 is opaque.

ggplot(penguins, aes(body_mass_g, bill_length_mm))+ 
     geom_point(aes(shape = species), size = 3, alpha = 0.6) 

Adding colour

Now lets explore the different species by adding colour with the code colour = species.

ggplot(penguins, aes(body_mass_g, bill_length_mm))+ 
     geom_point(aes(shape = species, colour = species), size = 3, alpha = 0.6) 

This red-green colour combination is colourblind unfrieldly, so lets change the colour of the points with scale_colour_manual. To ensure the shapes match with the names we will also use scale_shape_manual.

ggplot(penguins, aes(body_mass_g, bill_length_mm))+ 
     geom_point(aes(shape = species, colour = species), size = 3, alpha = 0.6)+ 
  
  scale_colour_manual(values = c("#C15CCB", "#00868B", "#FF6A00"), 
                      labels = c("Chinstrap", "Gentoo", "Adélie"))+
  scale_shape_manual(values = c(17, 15, 16),
                     labels = c("Chinstrap", "Gentoo", "Adélie"))

We won't change the points any more, so let's save the plot as penguin_plot, so we can build upon it.

penguin_plot <- ggplot(penguins, aes(body_mass_g, bill_length_mm))+ 
     geom_point(aes(shape = species, colour = species), size = 3, alpha = 0.6)+ 
  
     scale_colour_manual(values = c("#C15CCB", "#00868B", "#FF6A00"), 
                      labels = c("Chinstrap", "Gentoo", "Adélie"))+
     scale_shape_manual(values = c(17, 15, 16),
                     labels = c("Chinstrap", "Gentoo", "Adélie"))

Changing the background

You can change the background of ggplot2 figures in a variety of ways with:

• theme_gray()
• theme_bw()
• theme_linedraw()
• theme_light()
• theme_minimal()
• theme_classic()
• theme_void()
• theme_dark()

My personal favourite is theme_bw therefore we will continue to make our plot with this theme.

We will further remove the thicker lines in the background with panel.grid.major = element_blank(), and the thinner lines with panel.grid.minor = element_blank().

penguin_plot + 
  theme_bw()+ # set the background theme                                                 
  theme(panel.grid.major = element_blank(), # remove the major lines
        panel.grid.minor = element_blank()) # remove the minor lines

Changing the text size

There are several ways to change the size of the font, but we can quickly change all font size with theme_bw(base_size = 20).

penguin_plot <- penguin_plot + 
  theme_bw(base_size = 15)+                                                 
  theme(panel.grid.major = element_blank(),
        panel.grid.minor = element_blank()) 

penguin_plot

Renaming the axes and legend

You can rename the axes and legend using labs.

penguin_plot <- penguin_plot + 
  labs(x = "Body Mass (g)", y = "Bill Length (mm)", colour = "Species", shape = "Species")

penguin_plot

Change axes position

In the standard plots, the axes titles are really close to the plot. We will increase their distance with vjust:

penguin_plot <- penguin_plot + 
    theme(axis.title.y = element_text(vjust = 3))+ # increase distance from the y-axis
    theme(axis.title.x = element_text(vjust = -1)) # increase distance from the x-axis

penguin_plot

Change legend position

Legends can be positioned in a number of different ways using legend.position:

• theme(legend.position="top")
• theme(legend.position="bottom")
• theme(legend.position="left")
• theme(legend.position="right")
• theme(legend.position="none")

Legend loction can be also a numeric vector c(x,y), where x and y are the coordinates of the legend box. Their values should be between 0 - 1. c(0,0) is the “bottom left” and c(1,1) is the “top right” position.

penguin_plot <- penguin_plot + 
    theme(legend.position = c(0.85, 0.2))

penguin_plot

Saving the figure

Publications require high quality images and they specify the size and format required on their website.

ggsave

ggsave is one example of how to save your figures.

It allows you to save high quality images in a variety of different file types (e.g. "png", "eps", "ps", "tex", "pdf", "jpeg", "tiff", "png", "bmp", "svg", "wmf"). You can also specify the width and height in "in", "cm", or "mm", and specify the plot resolution with dpi.

# This will save the last plot for the code you ran:

ggsave("penguin_plot.pdf", 
       dpi = 600, 
       width = 100, height = 60, unit = "mm")

pdf

An alternative method is to use pdf. This allows you to specify the colour mode (e.g. cmyk).

pdf("ggplot-cmyk.pdf", width = 12 / 2.54, height = 8 / 2.54,
    colormodel = "cmyk")

print(penguin_plot)

dev.off()

More ggplot2 plots and tricks

Change the size of points based on other data

Size can be changed based on data within the penguins dataframe. For example, here we have changed the size of the points based on bill_depth_mm.

ggplot(penguins, aes(body_mass_g, bill_length_mm))+ 
     geom_point(aes(size = bill_depth_mm)) 

Adding an ellispe

Add 95% confidence interval ellispses with stat_ellipse.

ellipse<- penguin_plot + 
  stat_ellipse(aes(colour = species, level=0.95))
  

ellipse

Linear regression line

Add a linear regression line with geom_smooth(method=lm).

lm <- penguin_plot + 
  geom_smooth(method=lm, aes(colour = species))

lm 

facet_wrap

facet_wrap is a fantastic tool to slit plots based on a specified categorical column. Here we will use the species column.

penguin_plot + 
  facet_wrap(~species, ncol = 3, nrow = 1) + # specifying 3 columns, 1 row
  theme(legend.position = "none") # remove legend

Changing strip design

It is also possible to change the size, colour, face and fill colour of the facet strips with the following code:

penguin_plot + 
  facet_wrap(~species, ncol = 3, nrow = 1) + # specifying 3 columns, 1 row
  theme(legend.position = "none")+ # remove legend
    theme(strip.text.x = element_text(size = 16, color = "white", face = "bold"), 
          strip.background = element_rect(fill="black"))

Removing white space and free axes

We can also remove the free space using scales = free.

penguin_plot + 
  facet_wrap(~species, scales = "free")+ 
  theme(legend.position = "none")+ # remove legend
    theme(strip.text.x = element_text(size = 16, color = "white", face = "bold"), 
          strip.background = element_rect(fill="black"))

facet_grid: Free facet width

To allow the facets to be different widths, we must use facet_grid and space = "free". This can be used with scales = "free".

penguin_plot + 
  facet_grid(.~species, scales = "free", space = "free")  +
  theme(legend.position = "none")+ # remove legend
    theme(strip.text.x = element_text(size = 16, color = "white", face = "bold"), 
          strip.background = element_rect(fill="black"))

Bar plot

library(dplyr)
penguin_summary <-  penguins %>% 
                    group_by(species) %>% 
                    summarise(mean = mean(body_mass_g, na.rm = T), 
                              sd = sd(body_mass_g, na.rm = T))


ggplot(penguin_summary, aes(y = mean, x=species, fill = species)) + 
  geom_bar(stat="identity")+
   geom_errorbar(aes(ymin = mean-sd, ymax = mean+sd), 
                width=.1)  

Create the plot like the code above for the scatter plot.

bar <- ggplot(penguin_summary, aes(y = mean, x = species, fill = species)) + 
  geom_bar(stat = "identity")+
  geom_errorbar(aes(ymin = mean-sd, ymax = mean+sd), 
                width=.1)+
  theme_bw(base_size = 20)+
  theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank())+
  labs(x = "Species", y = "Body Mass (g)")+
 scale_fill_manual(values = c( "#FF6A00","#C15CCB",  "#00868B"))+
  theme(legend.position = "none") +
  scale_y_continuous(limits = c(0, 5700))+
  theme(axis.title.y = element_text(vjust = 3)) +
  theme(axis.title.x = element_text(vjust = -1)) 

bar

Histogram

ggplot(penguins, aes(x = flipper_length_mm, fill = species)) + 
  geom_histogram(alpha = 0.4)+
  scale_fill_manual(values = c( "#FF6A00","#C15CCB",  "#00868B"))

Density Plot

ggplot(penguins, aes(x = flipper_length_mm, fill = species)) + 
  geom_density(alpha = 0.4)+
  scale_fill_manual(values = c( "#FF6A00","#C15CCB",  "#00868B"))

 flipper <- ggplot(penguins, aes(x = flipper_length_mm, fill = species, colour = species)) + 
  geom_density(alpha = 0.4)+
  scale_fill_manual(values = c( "#FF6A00","#C15CCB",  "#00868B"))+
  scale_colour_manual(values = c( "#FF6A00","#C15CCB",  "#00868B"))+
  theme_bw(base_size = 20)+
  theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank())+
  labs(x = "Flipper Length (mm)", y = "Density", fill = "Species", colour = "Species")+
  #theme(legend.position = "none") +
  scale_x_continuous(expand = c(0,0), limits = c(165, 238))+
  scale_y_continuous(expand = c(0,0), limits = c(0, 0.065), breaks=seq(0, 0.065, 0.02))+
  theme(axis.title.y = element_text(vjust = 3)) +
  theme(axis.title.x = element_text(vjust = -1)) 

flipper 

Add a mean line

flipper_mean <- penguins %>%
        group_by(species) %>%
        summarise(mean = mean(flipper_length_mm, na.rm = TRUE))
  
density <- flipper +
  geom_vline(data = flipper_mean, aes(xintercept = mean, colour = species), linetype = "dashed", size = 1.3)

density

Box plot

ggplot(na.omit(penguins), aes(x=species, y=flipper_length_mm, fill=sex)) +
  geom_boxplot()

box_plot<- ggplot(na.omit(penguins), aes(x=species, y=flipper_length_mm, fill=sex)) +
  geom_boxplot()+
  theme_bw(base_size = 20)+
  theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank())+
  labs(x = "Species", y = "Flipper Length (mm)", fill = "Sex")+
   theme(axis.title.y = element_text(vjust = 3)) +
  theme(axis.title.x = element_text(vjust = -1)) +
  theme(legend.position = c(0.15, 0.8))

box_plot

Arrange plots

There are several different methods that allow you to arrange your plots into different panels.

ggarrange

Here we will arrange four of the plots that we made above into one figure using ggarrange. You can specify the number of rows nrow and ncol and add `labels.

Note we previously saved our figures (ellipse, density, bar, box_plot) in the Global Environment and we are calling on them here.

library(ggpubr)

ggarrange(ellipse, density, bar, box_plot,
          nrow = 2, ncol = 2,
          labels = c("a", "b", "c", "d"))

Same axes

If two plots have the same axes, you can remove one and use align = "v" to ensure the plots remain the same size.

# Remove y axes text and title
lm <- lm +
   theme(axis.title.y=element_blank(),
        axis.text.y=element_blank(),
        axis.ticks.y=element_blank())


ggarrange(ellipse, lm,
          nrow = 1, ncol = 2,
          labels = c("a", "b"),
          align = "v")

patchwork

patchwork is a very straighforward and intuative package for arranging plots. e.g:

• plot1 + plot2 aligns two plots next to each other
• plot1 / plot2 aligns plot2 under plot1

Lets make a fancy one.

We add the figure labels with e.g. labs(tag = 'a'). We will also make two of our plots half the size of the main plot with plot_layout(widths=c(2,1)).

library(patchwork)

(ellipse + labs(tag = 'a')| ((bar+ labs(tag = 'b')) / (box_plot +labs(tag = 'c')))) + plot_layout(widths=c(2,1))

Artwork by @CerrenRichards