Two main points:

1. Tweaking the look of the previous plots
2. Plot Resolution and Font Sizes

ggplot2 uses themes to control the overall appearance

ggplot(transport, aes(x = gender, y = height, fill = gender)) +
  geom_boxplot() +
  theme_bw()

My default is theme_bw()

• Removes the background grey, prints labels in black etc.

ggplot(transport, aes(x = gender, y = height, fill = gender)) +
  geom_boxplot() +
  theme_classic()

ggplot(transport, aes(x = gender, y = height, fill = gender)) +
  geom_boxplot() +
  theme_void()

You can set a default theme for a workspace or session

theme_set(theme_bw())

Now all plots in the workspace will use theme_bw()

The theme() function is also where you set:

• axis.text, legend attributes etc.
• Often uses elements to set an attribute

?theme

Changing text within themes uses element_text()

?element_text

ggplot(transport, aes(x = gender, y = height, fill = gender)) +
  geom_boxplot() +
  theme_bw() +
  theme(axis.text.x = element_text(
      angle = 90, hjust = 1, vjust = 0.5
  ))

Changing backgrounds and outlines uses element_rect()

ggplot(transport, aes(x = gender, y = height, fill = gender)) +
  geom_boxplot() +
  theme_bw() +
  theme(legend.background = element_rect(
      fill = "yellow", colour = "black"
  ))

To remove all attributes use element_blank() in place of element_rect(), element_text() or element_line()

ggplot(transport, aes(x = gender, y = height, fill = gender)) +
  geom_boxplot() +
  theme_bw() +
  theme(panel.grid = element_blank())

We can move the legend to multiple places:

ggplot(transport, aes(x = gender, y = height, fill = gender)) +
  geom_boxplot() +
  theme_bw() +
  theme(legend.position = "bottom")

Or we can use co-ordinates to place it inside the plotting region

ggplot(transport, aes(x = gender, y = height, fill = gender)) +
  geom_boxplot() +
  theme_bw() +
  theme(legend.position = c(0.85, 0.15))

The plotting region is assumed to have width and height of 1

We can also edit axes, fills, outlines etc. using scale_...() layers

ggplot(transport, aes(x = gender, y = height, fill = gender)) +
  geom_boxplot() +
  theme_bw() +
  scale_y_log10(limits = c(100, 200), 
                breaks = c(100, 125, 150, 175)) 

We can turn off or modify plot expansion

ggplot(transport, aes(x = gender, y = height, fill = gender)) +
  geom_boxplot() +
  theme_bw() +
  scale_y_continuous(expand = c(0, 0)) 

• The first a = 0 multiplies the scale by 1 + a
• The second b = 0 adds \(\pm\)b to the axis extrema

Alternatively, we can use expand_scale() to set expansion on either side.

ex <- expand_scale(mult = c(0, 0.05), add = c(0, 0))
ggplot(transport, aes(x = gender, y = height, fill = gender)) +
  geom_boxplot() +
  theme_bw() +
  scale_y_continuous(expand = ex) 

• The input to expand_scale() adjusts bottom/top or left/right depending on your axis.

ggplot(transport, aes(x = gender, y = height, fill = gender)) +
  geom_boxplot() +
  theme_bw() +
  scale_fill_manual(values = c("green", "blue"))

Colours can also be specified using hexadecimal codes

rgb(1, 0, 0)

[1] "#FF0000"

The main image formats are jpeg, png and tiff

• R can also export svg and pdf
• ggplot2 has the function ggsave()

?ggsave

The Plots Tab is the default graphics device

ggplot(transport, aes(x = gender, y = height, fill = gender)) +
  geom_boxplot() +
  theme_bw() +
  scale_fill_manual(values = c("green", "blue"))

ggsave("HeightByGender.png")

• ggsave defaults to 300dpi
• Defaults to the size of the Plots Tab

• Change the size manually by setting the width and height attributes

ggsave("HeightByGender.png", width = 18, height = 18,
       units = "cm")

• Getting the font size right can take ages
• Need to set correctly when we make the plot

• We initiate another graphics device using png(), jpeg(), pdf() etc.
• Turn the device off after creating the image using dev.off()
• Nothing will appear in the Plots tab

png("HeightByGender.png", width = 18, height = 18, 
    units = "cm", res = 300)
ggplot(transport, aes(x = gender, y = height, fill = gender)) +
  geom_boxplot() +
  theme_bw() +
  theme(text = element_text(size = 16)) +
  scale_fill_manual(values = c("green", "blue"))
dev.off()

Try to decide how big the plot will be in your final document

• Export using that size
• pdf and svg output are vector-based not pixel-based
• Scale very well with size

• Try loading and exploring the additional files:
• toothData, PCR and RealTimeData