Plotting with ggplot2

RaukR 2024 • Advanced R for Bioinformatics

Roy Francis

21-Jun-2024

Graphs

Graphing is an essential part of data analyses. Data with same summary statistics can look very different when plotted out.

Summary statistics do not always reflect how the data looks like.

Anscombe’s quartet comprises four data sets that have nearly identical simple descriptive statistics, yet have very different distributions and appear very different when graphed.

A more modern example dataset is datasaurus dozen. A set of 13 xy datasets that have nearly identical summary statistics but look very different when plotted out. One of the plots is in fact a dinosaur.

Anscombe’s quartet, Datasaurus

R graphics

Typical R plots with default settings in base R and ggplot2.

ggplot2 vs Base Graphics

hist(iris$Sepal.Length)

library(ggplot2)
ggplot(iris,aes(x=Sepal.Length))+
  geom_histogram()

For simple graphs, the base plot seems to take minimal coding effort compared to a ggplot graph.

ggplot2 vs Base Graphics

plot(iris$Petal.Length,iris$Petal.Width,
     col=c("red","green","blue")[iris$Species],
     pch=c(0,1,2)[iris$Species])
legend(x=1,y=2.5,
       legend=c("setosa","versicolor","virginica"),
       pch=c(0,1,2),col=c("red","green","blue"))

ggplot(iris,aes(Petal.Length,Sepal.Length,color=Species))+
  geom_point()

For anything beyond extremely basic plots, base plotting quickly become complex. More importantly, base plots do not have consistency in it’s functions or plotting strategy.

Why ggplot2?

• Consistent code for any type of plot (almost!)
• Flexible and modular (Add/remove components)
• Automatic legends, colors etc
• Save plot objects
• Themes for reusing styles
• Numerous add-ons/extensions
• Nearly complete structured graphing solution
• Adapted to other programming languages
  • Gadfly in Julia
  • gramm in MatLab
  • GGPlot in Perl
  • PlotNine , ggpy, lets-plot in Python

ggplot2 has a consistent logic and more structured code for plotting. There is bit of a learning curve, but once the code syntax and the logic is clear, it becomes easy to plot a huge variety of graphs.

Grammar Of Graphics

• Created by Hadley Wickham in 2005
• Data: Input data
• Geom: A geometry representing data. Points, Lines etc
• Aesthetic: Visual characteristics of the geometry. Size, Color, Shape etc
• Scale: How visual characteristics are converted to display values
• Statistics: Statistical transformations. Counts, Means etc
• Coordinates: Numeric system to determine position of geometry. Cartesian, Polar etc
• Facets: Split data into subsets

Traditional graphing tools generally have independent set of rules for different kinds of graphs and also labelled differently such as barplots, scatterplots, boxplots etc. Each graph has it’s own function and plotting strategy.

Leland Wilkinson’s The Grammar of Graphics introduces this idea that any kind of graph can be created by following a set of rules and put forward a framework that enables this.

Grammar of graphics (GOG) tries to unify all graphs under a common umbrella. GOG brings the idea that graphs are made up of discrete elements (data, aesthetics, geometry, statistics, coordinates, facets, themes etc) which can be mixed and matched to create any plot. This creates a consistent underlying framework to graphing.

ggplot (Grammar of graphics) was built in R by Hadley Wickham in 2005 as an implementation of Leland Wilkinson’s book Grammar of Graphics.

Building A Graph: Syntax

Building A Graph

data(iris)
ggplot(iris)

Building A Graph

ggplot(iris,aes(x=Sepal.Length,
                y=Sepal.Width))

Building A Graph

ggplot(iris,aes(x=Sepal.Length,
                y=Sepal.Width))+
  geom_point()

Building A Graph

ggplot(iris,aes(x=Sepal.Length,
                y=Sepal.Width,
                colour=Species))+
  geom_point()

Data • iris

• Input data is always an R data.frame object

Sepal.Length	Sepal.Width	Petal.Length	Petal.Width	Species
5.1	3.5	1.4	0.2	setosa
4.9	3.0	1.4	0.2	setosa
4.7	3.2	1.3	0.2	setosa
4.6	3.1	1.5	0.2	setosa
5.0	3.6	1.4	0.2	setosa
5.4	3.9	1.7	0.4	setosa

str(iris)

'data.frame':   150 obs. of  5 variables:
 $ Sepal.Length: num  5.1 4.9 4.7 4.6 5 5.4 4.6 5 4.4 4.9 ...
 $ Sepal.Width : num  3.5 3 3.2 3.1 3.6 3.9 3.4 3.4 2.9 3.1 ...
 $ Petal.Length: num  1.4 1.4 1.3 1.5 1.4 1.7 1.4 1.5 1.4 1.5 ...
 $ Petal.Width : num  0.2 0.2 0.2 0.2 0.2 0.4 0.3 0.2 0.2 0.1 ...
 $ Species     : Factor w/ 3 levels "setosa","versicolor",..: 1 1 1 1 1 1 1 1 1 1 ...

It’s a good idea to use str() to check the input dataframe to make sure that numbers are actually numbers and not characters, for example. Verify that factors are correctly assigned.

Data • diamonds

carat	cut	color	clarity	depth	table	price	x	y	z
0.23	Ideal	E	SI2	61.5	55	326	3.95	3.98	2.43
0.21	Premium	E	SI1	59.8	61	326	3.89	3.84	2.31
0.23	Good	E	VS1	56.9	65	327	4.05	4.07	2.31
0.29	Premium	I	VS2	62.4	58	334	4.20	4.23	2.63
0.31	Good	J	SI2	63.3	58	335	4.34	4.35	2.75
0.24	Very Good	J	VVS2	62.8	57	336	3.94	3.96	2.48

str(diamonds)

tibble [53,940 × 10] (S3: tbl_df/tbl/data.frame)
 $ carat  : num [1:53940] 0.23 0.21 0.23 0.29 0.31 0.24 0.24 0.26 0.22 0.23 ...
 $ cut    : Ord.factor w/ 5 levels "Fair"<"Good"<..: 5 4 2 4 2 3 3 3 1 3 ...
 $ color  : Ord.factor w/ 7 levels "D"<"E"<"F"<"G"<..: 2 2 2 6 7 7 6 5 2 5 ...
 $ clarity: Ord.factor w/ 8 levels "I1"<"SI2"<"SI1"<..: 2 3 5 4 2 6 7 3 4 5 ...
 $ depth  : num [1:53940] 61.5 59.8 56.9 62.4 63.3 62.8 62.3 61.9 65.1 59.4 ...
 $ table  : num [1:53940] 55 61 65 58 58 57 57 55 61 61 ...
 $ price  : int [1:53940] 326 326 327 334 335 336 336 337 337 338 ...
 $ x      : num [1:53940] 3.95 3.89 4.05 4.2 4.34 3.94 3.95 4.07 3.87 4 ...
 $ y      : num [1:53940] 3.98 3.84 4.07 4.23 4.35 3.96 3.98 4.11 3.78 4.05 ...
 $ z      : num [1:53940] 2.43 2.31 2.31 2.63 2.75 2.48 2.47 2.53 2.49 2.39 ...

R data.frame is a tabular format with rows and columns just like a spreadsheet.

Data • Format

Wide

Sepal.Length	Sepal.Width	Petal.Length	Petal.Width	Species
5.1	3.5	1.4	0.2	setosa
4.9	3.0	1.4	0.2	setosa
4.7	3.2	1.3	0.2	setosa

Long

Species	variable	value
setosa	Sepal.Length	5.1
setosa	Sepal.Length	4.9
setosa	Sepal.Length	4.7

The data must be cleaned up and prepared for plotting. The data must be ‘tidy’. Columns must be variables and rows must be observations. The data can then be in wide or long format depending on the variables to be plotted.

Geoms

geoms

p <- ggplot(iris)
# scatterplot
p+geom_point(aes(x=Sepal.Length,y=Sepal.Width))
# barplot
p+geom_bar(aes(x=Sepal.Length))
# boxplot
p+geom_boxplot(aes(x=Species,y=Sepal.Width))
# search
help.search("^geom_",package="ggplot2")

Geoms are the geometric components of a graph such as points, lines etc used to represent data. The same data can be visually represented in different geoms. For example, points or bars. Mandatory input requirements change depending on geoms.

Stats

• Stats compute new variables from input data.

• Geoms have default stats.

x <- ggplot(iris) + geom_bar(aes(x=Sepal.Length),stat="bin")
y <- ggplot(iris) + geom_bar(aes(x=Species),stat="count")
z <- ggplot(iris) + geom_bar(aes(x=Species,y=Sepal.Length),stat="identity")
wrap_plots(x,y,z,nrow=1)

Stats

• Plots can be built with stats.

x <- ggplot(iris) + stat_bin(aes(x=Sepal.Length),geom="bar")
y <- ggplot(iris) + stat_count(aes(x=Species),geom="bar")
z <- ggplot(iris) + stat_identity(aes(x=Species,y=Sepal.Length),geom="bar")
wrap_plots(x,y,z,nrow=1)

• Normally the data is plotted directly from input as it is.
• Some plots require the data to be computed or transformed. Eg. boxplot, histograms, smoothing, predictions, regression etc.

Stats

• Stats have default geoms.

plot	stat	geom
histogram	bin	bar
smooth	smooth	line
boxplot	boxplot	boxplot
density	density	line
freqpoly	freqpoly	line

Use args(geom_bar) to check arguments.

Aesthetics

• Aesthetic mapping

ggplot(iris)+
  geom_point(aes(x=Sepal.Length,
                 y=Sepal.Width,
                 size=Petal.Length,
                 alpha=Petal.Width,
                 shape=Species,
                 color=Species))

• Aesthetic parameter

ggplot(iris)+
  geom_point(aes(x=Sepal.Length,
                 y=Sepal.Width),
                 size=2,
                 alpha=0.8,
                 shape=15,
                 color="steelblue")

Aesthetics are used to assign values to geometries. For example, a set of points can be a fixed size or can be different colors or sizes denoting a variable.

This would be an incorrect way to do it.

ggplot(iris)+
geom_point(aes(x=Sepal.Length,y=Sepal.Width,size=2)

Aesthetics

x1 <- ggplot(iris) +
  geom_point(aes(x=Sepal.Length,y=Sepal.Width))+
  stat_smooth(aes(x=Sepal.Length,y=Sepal.Width))

x2 <- ggplot(iris,aes(x=Sepal.Length,y=Sepal.Width))+
                      geom_point() + geom_smooth()

wrap_plots(x1,x2,nrow=1,ncol=2)

If the same aesthetics are used in multiple geoms, they can be moved to ggplot().

Multiple Geoms

ggplot(iris,aes(x=Sepal.Length,y=Sepal.Width))+
      geom_point()+
      geom_line()+
      geom_smooth()+
      geom_rug()+
      geom_step()+
      geom_text(data=subset(iris,iris$Species=="setosa"),aes(label=Species))

Multiple geoms can be plotted one after the other. The order in which items are specified in the command dictates the plotting order on the actual plot.

In this case, the points appear over the lines.

ggplot(iris,aes(x=Sepal.Length,y=Sepal.Width))+
      geom_point()+
      geom_line()+

while here the lines appear above the points.

ggplot(iris,aes(x=Sepal.Length,y=Sepal.Width))+
      geom_line()+
      geom_point()+

Each geom takes input from ggplot() inputs. If extra input is required to a geom, it can be specified additionally inside aes().

data can be changed if needed for specific geoms.

Just because you can doesn’t mean you should!

Scales • Discrete Colors

• scales: position, color, fill, size, shape, alpha, linetype
• syntax: scale_<aesthetic>_<type>

p <- ggplot(iris)+geom_point(aes(x=Sepal.Length,
                     y=Sepal.Width,color=Species))
p

p + scale_color_manual(
     name="Manual",
     values=c("#5BC0EB","#FDE74C","#9BC53D"))

Scales are used to control the aesthetics. For example the aesthetic color is mapped to a variable x. The palette of colors used, the mapping of which color to which value, the upper and lower limit of the data and colors etc is controlled by scales.

Scales • Continuous Colors

• In RStudio, type scale_, then press TAB

p <- ggplot(iris)+
      geom_point(aes(x=Sepal.Length,
                     y=Sepal.Width,
      shape=Species,color=Petal.Length))
p

p +
scale_color_gradient(name="Pet Len",
  breaks=range(iris$Petal.Length),
  labels=c("Min","Max"),
  low="black",high="red")

Continuous colours can be changed using scale_color_gradient() for two colour gradient. Any number of breaks and colours can be specified using scale_color_gradientn().

Scales • Shape

p <- ggplot(iris)+
      geom_point(aes(x=Sepal.Length,
                     y=Sepal.Width,
      shape=Species,color=Species))
p

p +
scale_color_manual(name="New",
   values=c("blue","green","red"))+
scale_shape_manual(name="Bla",values=c(0,1,2))

Shape scale can be adjusted using scale_shape_manual(). Multiple mappings for the same variable groups legends.

Scales • Axes

• scales: x, y
• syntax: scale_<axis>_<type>
• arguments: name, limits, breaks, labels

p <- ggplot(iris)+geom_point(
  aes(x=Sepal.Length,y=Sepal.Width))
p

p + scale_x_continuous(name="Sepal Length",
        breaks=seq(1,8),limits=c(3,5))

The x and y axes are also controlled by scales. The axis break points, the break point text and limits are controlled through scales.

When setting limits using scale_, the data outside the limits are dropped. Limits can also be set using lims(x=c(3,5)) or xlim(c(3,5)). When mapping, coord_map() or coord_cartesian() is recommended for setting limits.

Facets • facet_wrap

• Split to subplots based on variable(s), Faceting in one dimension

p <- ggplot(iris)+
      geom_point(aes(x=Sepal.Length,
                     y=Sepal.Width,
                     color=Species))
p

p + facet_wrap(~Species)

p + facet_wrap(~Species,nrow=3)

facet_wrap is used to split a plot into subplots based on the categories in one or more variables.

Facets • facet_grid

• Faceting in two dimensions

p <- diamonds %>%
     ggplot(aes(carat,price))+
     geom_point()
p + facet_grid(~cut+clarity)

p + facet_grid(cut~clarity)

facet_grid is also used to split a plot into subplots based on the categories in one or more variables. facet_grid can be used to create a matrix-like grid of two variables.

Coordinate Systems

• coord_cartesian(xlim=c(2,8)) for zooming in
• coord_map for controlling limits on maps
• coord_polar for polar ccordinates

p <- ggplot(iris,aes(x="",y=Petal.Length,fill=Species))+
  geom_bar(stat="identity")
p

p + coord_polar("y", start = 0)

The coordinate system defines the surface used to represent numbers. Most plots use the cartesian coordinate sytem. Pie charts for example, is a polar coordinate projection of a cartesian barplot. Maps for example can have numerous coordinate systems called map projections. For example; UTM coordinates.

Theming

• Modify non-data plot elements/appearance
• Axis labels, panel colors, legend appearance etc

ggplot(iris, aes(Petal.Length)) +
    geom_histogram() +
    facet_wrap(~Species, nrow = 2) +
    theme_grey()

• Save a particular appearance for reuse
• ?theme

ggplot(iris, aes(Petal.Length)) +
    geom_histogram() +
    facet_wrap(~Species, nrow = 2) +
    theme_bw()

Themes allow to modify all non-data related components of the plot. This is the visual appearance of the plot. Examples include the axes line thickness, the background color or font family.

Theme • Legend

p <- ggplot(iris)+
      geom_point(aes(x=Sepal.Length,
                     y=Sepal.Width,
                     color=Species))

p + theme(legend.position="top")

p + theme(legend.position="bottom")

Theme • Text

element_text(family=NULL,face=NULL,color=NULL,size=NULL,hjust=NULL,
             vjust=NULL, angle=NULL,lineheight=NULL,margin = NULL)

p <- p + theme(
    axis.title=element_text(color="#e41a1c"),
    axis.text=element_text(color="#377eb8"),
    plot.title=element_text(color="#4daf4a"),
    plot.subtitle=element_text(color="#984ea3"),
    legend.text=element_text(color="#ff7f00"),
    legend.title=element_text(color="#ffff33"),
    strip.text=element_text(color="#a65628")
)

Theme • Rect

element_rect(fill=NULL,color=NULL,size=NULL,linetype=NULL)

p <- p + theme(
    plot.background=element_rect(fill="#b3e2cd"),
    panel.background=element_rect(fill="#fdcdac"),
    panel.border=element_rect(fill=NA,color="#cbd5e8",size=3),
    legend.background=element_rect(fill="#f4cae4"),
    legend.box.background=element_rect(fill="#e6f5c9"),
    strip.background=element_rect(fill="#fff2ae")
)

Theme • Reuse

newtheme <- theme_bw() + theme(
  axis.ticks=element_blank(), panel.background=element_rect(fill="white"),
  panel.grid.minor=element_blank(), panel.grid.major.x=element_blank(),
  panel.grid.major.y=element_line(size=0.3,color="grey90"), panel.border=element_blank(),
  legend.position="top", legend.justification="right"
)

p

p + newtheme

Professional themes

How BBC works with R graphics Visual design by Cedric Scherer

Position

us <- USArrests %>% mutate(state=rownames(.)) %>% slice(1:4) %>%
                    gather(key=type,value=value,-state)
p <- ggplot(us,aes(x=state,y=value,fill=type))

p + geom_bar(stat="identity",position="stack")

p + geom_bar(stat="identity",position="dodge")

Saving plots

p <- ggplot(iris,aes(Petal.Length,Sepal.Length,color=Species))+
  geom_point()

• ggplot2 plots can be saved just like base plots

png("plot.png",height=5,width=7,units="cm",res=200)
print(p)
dev.off()

• ggplot2 package offers a convenient function

ggsave("plot.png",p,height=5,width=7,units="cm",dpi=200,type="cairo")

• Note that default units in png is pixels while in ggsave it’s inches

Combining Plots

p <- ggplot(us,aes(x=state,y=value,color=type))+geom_point()
q <- ggplot(us,aes(x=state,y=value,fill=type))+geom_bar(stat="identity")

patchwork::wrap_plots(p,q)

Refer to patchwork documentation. Some notable alternatives are ggpubr and cowplot.

Combining two or more ggplot2 plots is often required and several packages exist to help with this situation. Some functions allow plots to be placed adjacently, also allowing varying heights or widths of each plot. Some functions allow one plot to be plotted on another plot like a subset plot. Here are alternative options.

gridExtra::grid.arrange(p,q,ncol=2)
ggpubr::ggarrange(p,q,ncol=2,widths=c(1.5,1),common.legend=T)
cowplot::plot_grid()

Interactive

• Convert ggplot2 object to interactive HTML

p <- ggplot(iris,aes(x=Sepal.Length,y=Sepal.Width,col=Species))

p1 <- p+geom_point()
plotly::ggplotly(p1,width=500,height=400)

p2 <- p+ggiraph::geom_point_interactive(
  aes(tooltip=paste0("<b>Species: </b>",Species)))+
  theme_bw(base_size=12)
ggiraph::ggiraph(code=print(p2))

Most interactive plotting libraries are not as complete as ggplot2. Therefore, some packages explore ways of converting ggplot2 objects into interactive graphics

Extensions

• ggrepel: Text labels including overlap control
• ggforce: Circles, splines, hulls, voronoi etc
• ggpmisc: Miscellaneaous features
• ggthemes: Set of extra themes
  
• ggthemr: More themes
  
• ggsci: Color palettes for scales
  
• ggmap: Dedicated to mapping
  
• ggraph: Network graphs
  
• ggiraph: Converting ggplot2 to interactive graphics

A collection of ggplot extension packages: https://exts.ggplot2.tidyverse.org/.
Curated list of ggplot2 links: https://github.com/erikgahner/awesome-ggplot2.

Learning & Help

• ggplot2 book
  
  
• The R cookbook
  

• ggplot2 official reference
  
• RStudio cheatsheet
  
• r-statistics ggplot2 cheatsheet
  
• StackOverflow
  
• Blogs, R-Bloggers, Cedric Scherer etc.

Bonus

ggcats

Bonus

ggdogs

Thank you! Questions?

         _                  
platform x86_64-pc-linux-gnu
os       linux-gnu          
major    4                  
minor    3.2                

2024 • SciLifeLab • NBIS • RaukR