1. Introduction
2. Rationale behind ggplot2
3. "Hello World" Plot
4. Basic Template
5. Themes
6. Aggregating Aesthetics
7. Line Plots
8. Faceting
9. Histograms
10. Scales
11. Box Plot
12. ggpubr + Multiple Plots Same Figure
13. Saving Plots

• Show Challenger space shuttle disaster (Edward R. Tufte book)

• Always plot the data.

• Replace (or complement) 'typical' tables of data or statistical results with figures that are more compelling and accessible.

• Whenever possible, generate figures that overlay / juxtapose observed data and analytical results, e.g. the 'fit'.

(source: jenny@UBC)

• Based on the Grammar of Graphics (Wilkinson, 2005)

• It can be difficult so Forget your preconceptions from other graphics tools

• ggplot2 is designed to work iteratively

• What is the grammar of graphics?

  • is an answer to a question: what is a statistical graphic?

1. data, in data.frame form
2. A set of aesthetic mappings between variables in the data and visual properties.. position, color, line type?, and
3. At least one geom to speficy what people see.. points? lines? bars?

• https://github.com/rstudio/cheatsheets/raw/master/data-visualization-2.1.pdf (ggplot2-cheatsheet)

Iris dataset gives measurements (cm) of sepal and petal from three species of Iris.

head(iris)

##   Sepal.Length Sepal.Width Petal.Length Petal.Width Species
## 1          5.1         3.5          1.4         0.2  setosa
## 2          4.9         3.0          1.4         0.2  setosa
## 3          4.7         3.2          1.3         0.2  setosa
## 4          4.6         3.1          1.5         0.2  setosa
## 5          5.0         3.6          1.4         0.2  setosa
## 6          5.4         3.9          1.7         0.4  setosa

We have our 1st essential component: Data, in data.frame form. Which aesthetics should we mapping to?

• Suppose we wish to plot in axis x and y repectively the sepal's length and width.
  • Mapping Sepal.Length and Sepal.Width to aesthetics x and y.

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

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

• Geometry: points (geom_point()).

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

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

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

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

• Specify the data inside the ggplot function.
• Anything else that goes in there becomes a global setting.
• Then add layers of geometric objects.

Do not worry about the main visual ! Use themes.

gp + theme_classic()

https://github.com/jrnold/ggthemes

https://cran.r-project.org/web/packages/ggsci/vignettes/ggsci.html

# install.packages("ggthemes")
library(ggthemes)
gp + theme_economist() + scale_colour_economist()

https://cran.r-project.org/web/packages/cowplot/index.html

# install.packages("cowplot")
library(cowplot)
gp

• To add aditional variables to a plot use other aesthetics like colour, shape, and size.

• These work in the same way as the x and y aesthetics.

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

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

• We could have mapped class to the size aesthetic in the same way.
• Do it!

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

## Warning: Using size for a discrete variable is not advised.

• We also could have mapped class to the shape aesthetic in the same way but increasing all the points size.
• Do it!

ggplot(data = iris, mapping = aes(x = Sepal.Length, y = Sepal.Width)) +
  geom_point(mapping = aes(shape = Species), size = 3)

ggplot(data = iris, mapping = aes(x = Sepal.Length, y = Sepal.Width)) +
  geom_point(mapping = aes(shape = Species), size = 3, position = "jitter")

• 25 built in shapes identified by numbers (+ scale_shape_manual(values=<VECTOR OF SHAPES>))
• What's the difference between squares 0, 15 and 22?
  • 0-14 border determined by colour
  • 15-18 filled with colour
  • 21-24 border of colour and filled with fill

Q1. What's gone wrong with this code? Why are the points not blue?

ggplot(data = iris, mapping = aes(x = Sepal.Length, y = Sepal.Width)) +
  geom_point(mapping = aes(colour = "blue"))

Q2. What happens if you map an aesthetic to something other than a variable name, like aes(colour = Sepal.Length > 7)?

ggplot(data = iris, mapping = aes(x = Sepal.Length, y = Sepal.Width)) +
  geom_point(mapping = aes(colour = Sepal.Length > 7))

Q3. Take a subset of diamonds dataset and reproduce the following plot:

d2 <- diamonds[sample(1:dim(diamonds)[1],1000),]

• A plot constructed with ggplot can have more than one geom
• Trends in data
• Our sepal width vs sepal length could use a linear regression

ggplot(data = iris, mapping = aes(x = Sepal.Length, y = Sepal.Width)) +
  geom_point(mapping = aes(colour = Species)) + geom_smooth(method = "lm")

ggplot(data = iris, mapping = aes(x = Sepal.Length, y = Sepal.Width)) +
  geom_point(mapping = aes(colour = Species)) +
    geom_smooth(method = "lm", aes(color = Species))

• Create separate graphs for subsets of data
• Two main functions:
  • facet_wrap(): define subsets as the levels of a single grouping variable
  • facet_grid(): define subsets as the crossing of two grouping variables

ggplot(data = iris, mapping = aes(x = Sepal.Length, y = Sepal.Width)) +
  geom_point(mapping = aes(colour = Species)) + geom_smooth(method = "lm", aes(color = Species)) + facet_wrap(~Species)

• Faceting along rows

ggplot(data = iris, mapping = aes(x = Sepal.Length, y = Sepal.Width)) +
  geom_point(mapping = aes(colour = Species)) + 
    geom_smooth(method = "lm", aes(color = Species)) + facet_grid(. ~ Species)

The following plot displays the total number of samples in the iris dataset grouped by Species

• x-axis displays species
• y-axis displays count (but count is not a var in iris dataset)

ggplot(data = iris) + 
    geom_bar(mapping = aes(x = Species, fill = Species))

• ggplot2 will compute new variables for our plot.

?geom_bar

• We have the number of occurrences of a variable in variable freq. This will be the y of our plot. Attention to the parameter stat. We change it to identity.

ggplot(data = iris_count) + 
    geom_bar(mapping = aes(x = Species, y=freq, fill=Species), stat = "identity")

ggplot(data = iris_count) + 
    geom_bar(mapping = aes(x = Species, y=freq, fill=Species), stat = "identity")

Q1. Construct a histogram of any continuous variables within iris dataset (tip: take a look at ggplot2 cheatsheet and play with parameter binwidth)

R1.

ggplot(data = iris) + geom_histogram(mapping = aes(x = Petal.Width), binwidth = 0.3)

• We can plug error bars into the chart. First let's tidy our data and calculate the mean and standard sample error for the length grouped by structure and Species.

gather_iris <- gather(iris, Sepal.Length:Petal.Width,
                             key="structure", value="size")
iris_sum <- gather_iris %>% group_by(structure, Species) %>%
                    summarise(m = mean(size), sd = sd(size), sem = sd(size)/(sqrt(n())))
iris_sum

## # A tibble: 12 x 5
## # Groups:   structure [?]
##    structure    Species        m    sd    sem
##    <chr>        <fct>      <dbl> <dbl>  <dbl>
##  1 Petal.Length setosa     1.46  0.174 0.0246
##  2 Petal.Length versicolor 4.26  0.470 0.0665
##  3 Petal.Length virginica  5.55  0.552 0.0780
##  4 Petal.Width  setosa     0.246 0.105 0.0149
##  5 Petal.Width  versicolor 1.33  0.198 0.0280
##  6 Petal.Width  virginica  2.03  0.275 0.0388
##  7 Sepal.Length setosa     5.01  0.352 0.0498
##  8 Sepal.Length versicolor 5.94  0.516 0.0730
##  9 Sepal.Length virginica  6.59  0.636 0.0899
## 10 Sepal.Width  setosa     3.43  0.379 0.0536
## 11 Sepal.Width  versicolor 2.77  0.314 0.0444
## 12 Sepal.Width  virginica  2.97  0.322 0.0456

So, to plot the mean with bars we need some data transformations.

bar_ann <- ggplot(data = iris_sum, mapping = aes(x = structure, y = m, fill = Species)) +
  geom_bar(position = "dodge", stat = "identity") +
  geom_errorbar(mapping = aes(ymin = (m-sd), ymax = (m+sd)),
                width = .2, position=position_dodge(.9)) +
    geom_errorbar(mapping = aes(ymin = (m-sem), ymax = (m+sem)), width = .2, position=position_dodge(0.9), colour='red') +
  xlab(NULL) + ylab("mean")
bar_ann

• Manually tunning your plots !
• The name is a tip for the function which a scale will apply (e.g., scale_x_log10(), scale_fill_grey(), scale_colour_brewer(), scale_y_continuous(), scale_y_discrete() )

p <- ggplot(data = iris) + 
  geom_point(mapping = aes(x = Sepal.Length, y = Sepal.Width, colour = Species)) 
p + scale_colour_brewer()

• http://colorbrewer2.org/

p + scale_colour_brewer(palette = "YlOrBr")

• Google it: hex colour picker

p + scale_colour_manual(values = c("#4286f4", "#1b4484", "#051a3a"))

Supply text strings (using \n for line breaks) or mathematical expressions in quote() (check in ?plotmath)

p + xlab("Sepal Length \n(cm)") + ylab("Sepal Width (cm)") + labs(colour = "Sp")

• Use the template + expression(paste(<STRING>, <MATH VAR>, <REMAINING STRING)). Remove axis label with NULL expression.

p + xlab("Sepal Length \n(cm)") + ylab(expression(paste("Sepal Width (",sqrt(cm),")"))) + 
  labs(colour = NULL) + scale_y_sqrt()

The breaks argument controls which values appear as tick marks on axes and keys on legends.

p + scale_x_continuous( breaks = seq(4, 8, 0.5) )

Each break can have an associated label, controlled by the labels argument.

p + scale_x_continuous( breaks = seq(4, 8, 0.5) , labels = paste(seq(4, 8, 0.5), "cm"))

• Use italics for species in plant names.

• Use box plots to illustrate the spread and differences of samples.
• \(n=20\) samples from \(N(0,1)\).

##     Min.  1st Qu.   Median     Mean  3rd Qu.     Max. 
## -3.03609 -0.59772  0.11208 -0.07907  0.56028  1.47123

Q1. Use the following transformation of iris data to plot a boxplot for all discreve variables (i.e. Sepal.Width, Sepal.Length, Petal.Width, Petal.Length) coloured by Species names.

gather_iris <- tidyr::gather(iris, Sepal.Length:Petal.Width,
                             key="structure", value="size")

R1.

ggplot(data = gather_iris) + 
  geom_boxplot(mapping = aes(x = structure, y = size, colour = Species)) +
  scale_x_discrete(limits = c("Petal.Width", "Sepal.Width", "Petal.Length", "Sepal.Length"))

Q2. With the same dataset construct a bar plot with the mean of each variable (tip: geom_bar()). Why it's not working?

#ggplot(data = gather_iris) + 
# geom_bar(mapping = aes(x = structure, y = size, colour = Species))

• Suppose we plot the box plot the Petal's length for Iris setosa and Iris virginica. We compare the mean between the two distributions as we can see bellow. How do we plot the significance?

## 
##  Welch Two Sample t-test
## 
## data:  iris[iris$Species == "setosa", "Petal.Length"] and iris[iris$Species == "virginica", "Petal.Length"]
## t = -49.986, df = 58.609, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -4.253749 -3.926251
## sample estimates:
## mean of x mean of y 
##     1.462     5.552

Adding text to a plot can be quite tricky. The main tool is geom_text() which adds labels at specified x and y positions.

1.Subset the data

iris_2sp <- iris[iris$Species %in% c("setosa", "virginica"), ]
head(iris_2sp)

##   Sepal.Length Sepal.Width Petal.Length Petal.Width Species
## 1          5.1         3.5          1.4         0.2  setosa
## 2          4.9         3.0          1.4         0.2  setosa
## 3          4.7         3.2          1.3         0.2  setosa
## 4          4.6         3.1          1.5         0.2  setosa
## 5          5.0         3.6          1.4         0.2  setosa
## 6          5.4         3.9          1.7         0.4  setosa

2.Plot and label

boxall <- ggplot(data = iris_2sp) + 
  geom_boxplot(mapping = aes(x = Species, y = Petal.Length)) +
  ylim(c(0, 10)) + geom_segment(x = 1, xend = 2, y = 8, yend=8) +
  geom_text(x = 1.5, y = 9, label = "***", family = "mono", size=10)
boxall

Features:

• automatically add p-vlues and significance levels to box plots, bar plots and more.
• arrange and annotate multiple plots on the same page.
• change graphical parameters such as colors and labels.

library(ggpubr)
my_comparisons = list( c("setosa", "virginica") )
p <- ggboxplot(iris_2sp, x = "Species", y="Petal.Length", color="Species",
               add="jitter", shape="Species") +
        stat_compare_means(comparisons = my_comparisons, 
                           label = "p.signif", method = "t.test") 
p

plabeled <- ggboxplot(iris_2sp, x = "Species", y="Petal.Length", color="Species", 
                      add="jitter", shape="Species") +
        stat_compare_means(comparisons = my_comparisons, method = "t.test",
                           label="p.format", label.y = 8) 
plabeled

ggbarplot(gather_iris, x = "structure", y = "size", color = "Species", 
          add=c("mean_sd"), palette = "jco", position = position_dodge(0.9), ylab="mean")

• If the plot is on your screen

ggsave("~/path/to/figure/filename.png")

• If your plot is assigned to an object

ggsave(plot1, file = "~/path/to/figure/filename.png")

• Specify a size

ggsave(plot1, file = "~/path/to/figure/filename.png", width = 6,
        height = 4)

• Or any format (pdf, png, eps, svg, jpg)

ggsave("~/path/to/figure/filename.eps")
ggsave("~/path/to/figure/filename.svg")
ggsave("~/path/to/figure/filename.pdf")

library(cowplot)

ggdraw()

ggdraw() + draw_plot(boxall, x = 0, y = 0, 
                     width = 0.5, height = 1) +
           draw_plot(p, x = 0.5, y = 0,
                     width = 0.5, height = 1)

library(RColorBrewer)

pal <- brewer.pal(name = "RdYlBu", n = 3)

pmulti <- ggdraw() +
    draw_plot(bar_ann + scale_fill_manual(values = pal) + 
                  theme(axis.text.x = element_text(angle=60, vjust=0.5), 
                        plot.margin = unit(c(6,0,6,0), "pt"), legend.position = "none") + 
                  ylab(""), x = 0, y = 0.1, width = 0.5, height = 0.8) 

pmulti <- pmulti + draw_plot(p + scale_colour_manual(values = c(pal[1], pal[3])) + 
                  theme(plot.margin = unit(c(6,6,6,6), "pt")) + 
                  theme(legend.position = "none"), x = 0.5, y = 0.2,
                    width = 0.25, height = 0.8) +
    draw_plot(plabeled + scale_colour_manual(values = c(pal[1], pal[3])) +
                  theme(plot.margin = unit(c(6,6,6,6), "pt"))+ 
                  theme(legend.position = "none"), x = 0.75, y = 0.2,
                    width = 0.25, height = 0.8)

# write plot labels (A, B, C ...)
pmulti <- pmulti + draw_plot_label(label = c("A", "B", "C"), 
                                   size = 15, x = c(0, 0.5, 0.75), y =c(1, 1, 1))

# extract the legend from one of the plots
# (clearly the whole thing only makes sense if all plots
# have the same legend, so we can arbitrarily pick one.)
legend <- get_legend(bar_ann + theme(legend.position = "bottom") + scale_fill_manual(values = pal))
pmulti <- pmulti + draw_plot(legend, x=0, y=0)

#save_plot("/Users/rodtheo/Downloads/temp/firstDeck/figure/multiplot.png", 
#               pmulti, base_height = 8, base_width=12)