G - 5 Data Visualization with ggplot2

Overview

Teaching: 45 min
Exercises: 30 min

Questions

• How can I create simple plots with ggplot?

• What is faceting in ggplot?

• How can I change the aestetics (e.g. axis labels and color) of a plot?

Objectives

• Produce scatter plots, boxplots, and barplots using ggplot.

• Set universal plot settings.

• Describe what faceting is and apply faceting in ggplot.

• Modify the aesthetics of an existing ggplot plot (including axis labels and color).

• Build complex and customized plots from data in a data frame.

Getting ready for plotting

We start by loading the required packages. ggplot2 is included in the tidyverse package.

library(tidyverse)

If not still in the workspace, load the samples dataset used in the previous episode.

samples <- read_csv("data_raw/covid_samples.csv")

Like previously, we turn the disease_outcome and sex columns into factors:

samples$disease_outcome <- factor(samples$disease_outcome)
samples$sex <- factor(samples$sex)

Loading the sequencing data

To make plotting a bit more interesting, we will in this episode also make use of the COVseq-MiSeq dataset mentioned in episode 3. This dataset contains results from the sequencing of the SARS-CoV-2 samples using the COVseq method on a MiSeq instrument from Illumina. We will first load the sequencing data into a dataframe, and then combine that data frame with the samples data frame.

A MiSeq instrument from Illumina.

The sequencing dataset has the following columns:

Let’s read the data into a new data frame named sequencing:

sequencing <- read_csv("data_raw/covseq_miseq.csv")

Now that we have created the data frame, we are going to transform two of the columns into factors:

sequencing$pangolin_lineage <- factor(sequencing$pangolin_lineage)
sequencing$nextclade_clade <- factor(sequencing$nextclade_clade)

Joining the sequencing and samples data

We now have two separate, but obviously related data frames. Both the samples and sequencing data frames have rows that correspond to the 29 SARS-CoV-2 samples. They both also have and a column named patient_id that uniquely identifies each sample. By joining the two data frames based on the patient ID, we can ensure that each row in the sequencing data frame is paired with the correct row in the samples data frame.

Joining two related datasets based on shared variables is fairly common in data analysis workflows. We will use the dplyr function inner_join() to accommodate this in R:

covseq <- sequencing %>%
  inner_join(samples, by = "patient_id")

Note

dplyr offers 4 “mutating joins” functions for different situations: inner_join(), left_join(), right_join() and full_join. You can read more about the functions in R:s built-in help and here Links to an external site..

We should now have a data frame named covseq with 29 rows and all the columns from the sequencing and samples data frames. We can check this with the str() function:

str(covseq)

spec_tbl_df [29 × 20] (S3: spec_tbl_df/tbl_df/tbl/data.frame)
 $ alias           : chr [1:29] "PE300_COVseq_OAS-1" "PE300_COVseq_OAS-10" "PE300_COVseq_OAS-11" "PE300_COVseq_OAS-12" ...
 $ patient_id      : chr [1:29] "OAS-29_1" "OAS-29_10" "OAS-29_11" "OAS-29_12" ...
 $ instrument_model: chr [1:29] "Illumina MiSeq" "Illumina MiSeq" "Illumina MiSeq" "Illumina MiSeq" ...
 $ library_name    : chr [1:29] "COVseq" "COVseq" "COVseq" "COVseq" ...
 $ total_reads     : num [1:29] 46280 605294 918492 853598 775350 ...
 $ mapped_reads    : num [1:29] 46260 605266 918430 853536 775298 ...
 $ mapped_reads_pct: num [1:29] 100 100 100 100 100 ...
 $ coverage_median : num [1:29] 22 1211 1530 1514 1567 ...
 $ coverage_pct_1x : num [1:29] 96 100 100 100 100 100 99 95 98 100 ...
 $ coverage_pct_10x: num [1:29] 77 98 99 98 99 98 97 73 92 98 ...
 $ snps            : num [1:29] 14 15 6 17 16 14 18 16 15 15 ...
 $ pangolin_lineage: Factor w/ 3 levels "B.1","B.1.1.1",..: 2 2 3 2 2 2 2 1 3 2 ...
 $ nextclade_clade : Factor w/ 2 levels "20A","20D": 2 2 1 2 2 2 2 1 1 2 ...
 $ collection_date : Date[1:29], format: "2020-03-31" "2020-03-31" ...
 $ country         : chr [1:29] "Italy" "Italy" "Italy" "Italy" ...
 $ region          : chr [1:29] "Turin" "Turin" "Turin" "Turin" ...
 $ age             : num [1:29] 48 35 59 60 83 21 44 55 81 63 ...
 $ disease_outcome : Factor w/ 2 levels "dead","recovered": 1 NA 2 2 1 1 2 2 1 2 ...
 $ sex             : Factor w/ 2 levels "female","male": 1 2 2 1 1 2 1 2 1 1 ...
 $ ct              : num [1:29] 41.5 15.3 25.3 27 25.3 ...
 - attr(*, "spec")=
  .. cols(
  ..   alias = col_character(),
  ..   patient_id = col_character(),
  ..   instrument_model = col_character(),
  ..   library_name = col_character(),
  ..   total_reads = col_double(),
  ..   mapped_reads = col_double(),
  ..   mapped_reads_pct = col_double(),
  ..   coverage_median = col_double(),
  ..   coverage_pct_1x = col_double(),
  ..   coverage_pct_10x = col_double(),
  ..   snps = col_double(),
  ..   pangolin_lineage = col_character(),
  ..   nextclade_clade = col_character()
  .. )
 - attr(*, "problems")=<externalptr> 

Plotting with ggplot2

ggplot2 is a plotting package that makes it simple to create complex plots from data in a data frame. It provides a more programmatic interface for specifying what variables to plot, how they are displayed, and general visual properties. Therefore, we only need minimal changes if the underlying data change or if we decide to change from a bar plot to a scatterplot. This helps in creating publication quality plots with minimal amounts of adjustments and tweaking.

ggplot2 plots work best with data in the ‘long’ format, i.e. a column for every dimension, and a row for every observation. Well-structured data will save you lots of time when making figures with ggplot2.

ggplot graphics are built layer by layer by adding new elements. Adding layers in this fashion allows for extensive flexibility and customization of plots.

To build a ggplot, we will use the following basic template that can be used for different types of plots:

ggplot(data = <DATA>, mapping = aes(<MAPPINGS>)) +  <GEOM_FUNCTION>()

• use the ggplot() function and bind the plot to a specific data frame using the data argument:

ggplot(data = covseq)

• define an aesthetic mapping (using the aesthetic (aes) function), by selecting the variables to be plotted and specifying how to present them in the graph, e.g. as x/y positions, or characteristics such as size, shape, color, etc:

ggplot(data = covseq, mapping = aes(x = ct, y = total_reads))

• add ‘geoms’ (geometric objects) – graphical representations of the data in the plot (points, lines, bars). ggplot2 offers many different geoms; we will use some common ones today, including:

  • geom_point() for scatter plots, dot plots, etc.
  • geom_boxplot() for, well, boxplots!
  • geom_bar() for displaying relations between numeric and categorical variables.

To add a geom to the plot use + operator. Let’s first try geom_point():

ggplot(data = covseq, mapping = aes(x = total_reads, y = coverage_pct_10x)) +
  geom_point()

The + in the ggplot2 package is particularly useful because it allows you to modify existing ggplot objects. This means you can easily set up plot “templates” and conveniently explore different types of plots, so the above plot can also be generated with code like this:

# Assign plot to a variable
reads_10x_plot <- ggplot(data = covseq,
                         mapping = aes(x = total_reads, y = coverage_pct_10x))

# Draw the plot
reads_10x_plot + 
  geom_point()

Notes

• Anything you put in the ggplot() function can be seen by any geom layers that you add (i.e. these are universal plot settings). This includes the x- and y-axis you set up in aes().
• You can also specify aesthetics for a given geom independently of the aesthetics defined globally in the ggplot() function.
• The + sign used to add layers must be placed at the end of each line containing a layer. If, instead, the + sign is added in the line before the other layer, ggplot2 will not add the new layer and will return an error message.

# This is the correct syntax for adding layers
reads_10x_plot + 
  geom_point()

# This will not add the new layer and will return an error message
reads_10x_plot
  + geom_point()

Challenge 5.1

Create a scatter plot with the Ct value plotted against the total number of sequence reads.

Solution

ggplot(data = covseq, 
       mapping = aes(x = ct, y = total_reads)) +
  geom_point()

Building your plots iteratively

Building plots with ggplot2 is typically an iterative process. We start by defining the dataset we’ll use, lay out the axes, and choose a geom:

ggplot(data = covseq, mapping = aes(x = total_reads, y = coverage_pct_10x)) +
  geom_point()

Then, we start modifying this plot to extract more information from it. For instance, we can add transparency (alpha) to avoid overplotting:

ggplot(data = covseq, mapping = aes(x = total_reads, y = coverage_pct_10x)) +
  geom_point(alpha = 0.5)

We can also add colors for all the points:

ggplot(data = covseq, mapping = aes(x = total_reads, y = coverage_pct_10x)) +
    geom_point(alpha = 0.5, color = "blue")

Or to color each point in the plot differently, you could use a vector as an input to the argument color. ggplot2 will provide a different color corresponding to different values in the vector. Here is an example where we color with disease_outcome:

ggplot(data = covseq, mapping = aes(x = total_reads, y =coverage_pct_10x)) +
    geom_point(alpha = 0.5, aes(color = disease_outcome))

Challenge 5.2

Use what you just learned to create a scatter plot of the sex against the cycle threshold (Ct). Is this a good way to show this type of data?

Solution

ggplot(data = covseq, 
       mapping = aes(x = sex, y = ct)) +
  geom_point(aes(color = sex))

Boxplot

Another useful way to visualize and compare distributions across groups is the boxplot. Here we will first create a boxplot that visualizes the distribution of Ct values within each sex:

ggplot(data = covseq, mapping = aes(x = sex, y = ct)) +
    geom_boxplot()

By adding points to the boxplot, we can have a better idea of the number of counts and of their distribution:

ggplot(data = covseq, mapping = aes(x = sex, y = ct)) +
    geom_boxplot(alpha = 0) +
    geom_jitter(alpha = 0.5, color = "tomato")

Notice how the boxplot layer is behind the jitter layer? What do you need to change in the code to put the boxplot in front of the points such that it’s not hidden?

Challenges 5.3

Boxplots are useful summaries, but hide the shape of the distribution. For example, if there is a bimodal distribution, it would not be observed with a boxplot. An alternative to the boxplot is the violin plot (sometimes known as a beanplot), where the shape (of the density of points) is drawn. Replace the box plot with a violin plot; see geom_violin(). Modify the code below to show a violin plot instead.

ggplot(data = covseq, mapping = aes(x = sex, y = ct)) +
    geom_boxplot(alpha = 0) +
    geom_jitter(alpha = 0.5, color = "tomato")

Solution

ggplot(data = covseq, mapping = aes(x = sex, y = ct)) +
    geom_violin(alpha = 0) +
    geom_jitter(alpha = 0.5, color = "tomato")

In many types of data, it is important to consider the scale of the observations. For example, it may be worth changing the scale of the axis to better distribute the observations in the space of the plot. Changing the scale of the axes is done similarly to adding/modifying other components.

• Modify the code below so that the number of reads are shown on a log 10 scale; see scale_x_log10().

ggplot(data = covseq, mapping = aes(x = total_reads, y = coverage_pct_10x)) +
    geom_point(alpha = 0)

Solution

ggplot(data = covseq, mapping = aes(x = total_reads, y = coverage_pct_10x)) +
  geom_point() +
  scale_x_log10()

• Add color to the data points on your plot according to the disease outcome.

Solution

ggplot(data = covseq, mapping = aes(x = total_reads, y = coverage_pct_10x)) +
  geom_point(aes(color = disease_outcome)) +
  scale_y_log10()

• Replace “NA” with “unknown” (hint: use the addNA() and levels() functions).

Solution

# Add NAs to the factor
covseq$disease_outcome <- addNA(covseq$disease_outcome)

# Rename the level
levels(covseq$disease_outcome)[3] <- "unknown"

# Now create the plot in the same way as before
ggplot(data = covseq, mapping = aes(x = total_reads, y = coverage_pct_10x)) +
  geom_point(aes(color = disease_outcome)) +
  scale_y_log10()

Barplot

Another common type of plot is the barplot. This kind of plot can be created with geom_bar(). In order to create a barplot, we will first prepare a suitable dataset:

# Count the 
sex_cnts <- covseq %>% count(sex)

sex_cnts

# A tibble: 2 × 2
  sex        n
  <fct>  <int>
1 female    16
2 male      13

Let’s then create a barplot from the tiny dataset that we just created:

ggplot(sex_cnts, aes(x = sex, y = n)) +
  geom_bar(stat = "identity")

In the code above, we used the argument stat = "identity" instead of the default value bin. This means that the height of the bar will be represented by the count in each category.

We can improve the plot by using different fill colors for the sexes:

ggplot(sex_cnts, aes(x = sex, y = n, fill = sex)) +
  geom_bar(stat = "identity")

We could also have added this configuration to the geom_bar() layer instead:

ggplot(sex_cnts, aes(x = sex, y = n)) +
  geom_bar(stat = "identity", aes(fill = sex))

Integrating the pipe operator with ggplot2

In the previous lesson, we saw how to use the pipe operator %>% to use different functions in a sequence and create a coherent workflow. We can also use the pipe operator to pass the data argument to the ggplot() function. The hard part is to remember that to build your ggplot, you need to use + and not %>%.

sex_cnts %>% ggplot(aes(x = sex, y = n, fill = sex)) +
  geom_bar(stat = "identity")

The pipe operator can also be used to link data manipulation with consequent data visualization.

sex_cnts_plot <- covseq %>%
  count(sex) %>%
  ggplot(aes(x = sex, y = n, fill = sex)) +
  geom_bar(stat = "identity")

sex_cnts_plot

Faceting

ggplot has a special technique called faceting that allows the user to split one plot into multiple plots based on a factor included in the dataset. We will use it to make one barplot for each of disease outcome:

covseq %>%
  # Count sex and disease outcome
  count(sex, disease_outcome) %>%
  # Create a separate barplot for each disease outcome
  ggplot(aes(x = sex, y = n, fill = sex)) +
    geom_bar(stat = "identity") +
    facet_wrap(facets = vars(disease_outcome))

You can also create more advanced layouts using the facet_grid() function. We can for example arrange the plots vertically instead of horizontally:

covseq %>%
  # Count sex and disease outcome
  count(sex, disease_outcome) %>%
  # Create a separate barplot for each disease outcome
    ggplot(aes(x = sex, y = n, fill = sex)) +
    geom_bar(stat = "identity") +
    facet_grid(rows = vars(disease_outcome))

Note: ggplot2 before version 3.0.0 used formulas to specify how plots are faceted. If you encounter facet_grid/wrap(...) code containing ~, please read https://ggplot2.tidyverse.org/news/#tidy-evaluation Links to an external site..

ggplot2 themes

Usually plots with white background look more readable when printed. Every single component of a ggplot graph can be customized using the generic theme() function, as we will see below. However, there are pre-loaded themes available that change the overall appearance of the graph without much effort.

For example, we can change our previous graph to have a simpler white background using the theme_bw() function:

covseq %>%
  # Count sex and disease outcome
  count(sex, disease_outcome) %>%
  # Create a separate barplot for each disease outcome
  ggplot(aes(x = sex, y = n, fill = sex)) +
    geom_bar(stat = "identity") +
    facet_wrap(facets = vars(disease_outcome)) +
    theme_bw()

In addition to theme_bw(), which changes the plot background to white, ggplot2 comes with several other themes which can be useful to quickly change the look of your visualization. The complete list of themes is available at https://ggplot2.tidyverse.org/reference/ggtheme.html Links to an external site.. theme_minimal() and theme_light() are popular, and theme_void() can be useful as a starting point to create a new hand-crafted theme.

The ggthemes Links to an external site. package provides a wide variety of options.

Challenge 5.4

Use what you just learned to create a plot that shows how counts of PANGO lineages (pangolin_lineage) differ between disease outcomes.

Solution

covseq %>%
  count(pangolin_lineage, disease_outcome) %>%
  ggplot(aes(x = disease_outcome, y = n, fill = disease_outcome)) +
    geom_bar(stat = "identity") +
    facet_wrap(vars(pangolin_lineage))

Customization

Take a look at the ggplot2 cheat sheet Links to an external site., and think of ways you could improve the plot.

Now, let’s start with changing the names of axes and add a title to the figure:

covseq %>%
  # Count sex and disease outcome
  count(sex, disease_outcome) %>%
  # Create a separate barplot for each disease outcome
  ggplot(aes(x = sex, y = n, fill = sex)) +
    geom_bar(stat = "identity") +
    facet_wrap(facets = vars(disease_outcome)) +
    labs(title = "Females and males per disease outcome",
         x = "Sex",
         y = "Number of patients") +
    theme_bw()

The axes have more informative names, but their readability can be improved by increasing the font size. This can be done with the generic theme() function:

covseq %>%
  # Count sex and disease outcome
  count(sex, disease_outcome) %>%
  # Create a separate barplot for each disease outcome
  ggplot(aes(x = sex, y = n, fill = sex)) +
    geom_bar(stat = "identity") +
    facet_wrap(facets = vars(disease_outcome)) +
    labs(title = "Females and males per disease outcome",
         x = "Sex",
         y = "Number of patients") +
    theme(text = element_text(size = 16))  # set the font size of text elements

We can also add another layer with the scale_fill_discrete() function to adjust the figure legend:

covseq %>%
  # Count sex and disease outcome
  count(sex, disease_outcome) %>%
  # Create a separate barplot for each disease outcome
  ggplot(aes(x = sex, y = n, fill = sex)) +
    geom_bar(stat = "identity") +
    facet_wrap(facets = vars(disease_outcome)) +
    labs(title = "Females and males per disease outcome",
         x = "Sex",
         y = "Number of patients") +
    scale_fill_discrete(
      name = "Sex",  # figure legend title
      labels = c("Female", "Male")) +  # new figure legend labels
    theme_bw() +
    theme(text = element_text(size = 16))

Note that it is also possible to change the fonts of your plots. If you are on Windows, you may have to install the extrafont package Links to an external site., and follow the instructions included in the README for this package.

Challenge 5.5

With all of this information in hand, please take another five minutes to either improve one of the plots generated in this exercise or create a beautiful graph of your own. Use the RStudio ggplot2 cheat sheet Links to an external site. for inspiration.

Here are some ideas:

• See if you can change the thickness of the lines.
• Can you find a way to change the name of the legend? What about its labels?
• Try using a different color palette (see http://www.cookbook-r.com/Graphs/Colors_(ggplot2)/ Links to an external site.).

Arranging plots

Faceting is a great tool for splitting one plot into multiple plots, but sometimes you may want to produce a single figure that contains multiple plots using different variables or even different data frames. We won’t go into it here, but the * patchwork package can be used to combine separate gplots into a single figure while keeping everything aligned properly. Like most R packages, patchwork can be installed from CRAN, the R package repository.

Exporting plots

After creating your plot, you can save it to a file in your favorite format. The Export tab in the Plot pane in RStudio will save your plots at low resolution, which will not be accepted by many journals and will not scale well for posters. The ggplot2 extensions website Links to an external site. provides a list of packages that extend the capabilities of ggplot2, including additional themes.

Instead, use the ggsave() function, which allows you easily change the dimension and resolution of your plot by adjusting the appropriate arguments (width, height and dpi):

sex_outcome_plot <- covseq %>%
  # Count sex and disease outcome
  count(sex, disease_outcome) %>%
  # Create a separate barplot for each disease outcome
  ggplot(aes(x = sex, y = n, fill = sex)) +
    geom_bar(stat = "identity") +
    facet_wrap(facets = vars(disease_outcome)) +
    labs(title = "Females and males per disease outcome",
         x = "Sex",
         y = "Number of patients") +
    scale_fill_discrete(
      name = "Sex",  # figure legend title
      labels = c("Female", "Male")) +  # new figure legend labels
    theme_bw() +
    theme(text = element_text(size = 16))
  
# Save the file in a subdirectory named "fig"
ggsave("fig/sex_disease_outcome.png", sex_outcome_plot, width = 15, height = 10)