In AshirBorah/cp_bootcamp_r_tutorials:

library(learnr)
library(gapminder)
library(ggrepel)
library(tidyverse)
library(plotly)


knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE, fig.width = 5, fig.height = 3.5)
tutorial_options(exercise.timelimit = 20, exercise.blanks = "___+")


gapminder <- gapminder %>% 
  mutate(country = as.character(country),
         continent = as.character(continent))

gap_92 <- gapminder %>% 
  filter(year == 1992) %>% 
  mutate(gdp = pop*gdpPercap/1e9)

Scatterplot basics {data-progressive=TRUE}

Create data

We'll be using the gapminder data again

head(gapminder)

First, pull out the gapminder data from the year 1992, and add a column gdp (in units of billions of $). Call this new table gap_92

gap_92 <- ___

gap_92 <- gapminder %>% 
  filter(year == 1992) %>% 
  mutate(gdp = pop*gdpPercap/1e9)

Recreate Graph 1

For the following you will be able to use the gap_92 table you created above. Let's make a basic scatterplot of the data using ggplot.

Your output should look like this.

ggplot(data = gap_92, mapping = aes(x = gdp, y = lifeExp)) + 
  geom_point()

gap_92 <- gapminder %>% 
  filter(year == 1992) %>% 
  mutate(gdp = pop*gdpPercap/1e9)

ggplot(data = gap_92, mapping = aes(x = gdp, y = lifeExp)) + 
  geom_point()

Recreate Graph 2

Now make the same figure, but set the x-axis on a log10 scale, add informative x- and y-axis labels, and a figure title.

Your new plot should look like this:

ggplot(data = gap_92, mapping = aes(x = gdp, y = lifeExp)) + 
  geom_point() +
  scale_x_log10() +
  labs(x = 'GDP (billions $)', y = 'Life expectancy', title = 'Life Expectancy vs GDP')

ggplot(data = gap_92, mapping = aes(x = gdp, y = lifeExp)) + 
  geom_point() +
  scale_x_log10() +
  labs(x = 'GDP (billions $)', y = 'Life expectancy', title = 'Life Expectancy vs GDP')

Additional Practice

Now explore mapping additional information from this dataset, such as continent and pop to different aesthetic variables of the points. Use help(geom_point) to see information on the available aesthetics for points.

Line plots and color basics

Line plot base

Pick 3 countries and make a plot of their lifeExp over time including both lines and points. Make the lines colored by country, but make the points all black.

Your plot should look similar to this.

target_countries <- c('Venezuela', 'Vietnam', 'Uruguay')
gap_targ <- gapminder %>% 
  filter(country %in% target_countries)
ggplot(gap_targ, mapping = aes(x = year, lifeExp, color = country)) + 
  geom_line() + 
  geom_point(color = 'black')

## One of the solution
target_countries <- c('Venezuela', 'Vietnam', 'Uruguay')
gap_targ <- gapminder %>% 
  filter(country %in% target_countries)
ggplot(gap_targ, mapping = aes(x = year, lifeExp, color = country)) + 
  geom_line() + 
  geom_point(color = 'black')

Transparency and size

Try controlling the size and transparency of the points. (transparancy is controlled by 'alpha', which ranges from 0 to 1)

## One of the solution
target_countries <- c('Venezuela', 'Vietnam', 'Uruguay')
gap_targ <- gapminder %>% 
  filter(country %in% target_countries)
ggplot(gap_targ, mapping = aes(x = year, lifeExp, color = country)) + 
  geom_line() + 
  geom_point(color = 'black', size=3, alpha=0.4)

Use scale_color_manual to set the colors to ones you like. You can choose specific colors for each country using a named vector of colors like this: my_colors = c(country1 = 'darkred', country2 = 'magenta', ...). You can see names of available colors here.

my_colors = c(Venezuela = 'darkred', Vietnam = 'magenta', Uruguay='blue')

target_countries <- c('Venezuela', 'Vietnam', 'Uruguay')
gap_targ <- gapminder %>% 
  filter(country %in% target_countries)
ggplot(gap_targ, mapping = aes(x = year, lifeExp, color = country)) + 
  geom_line() + 
  geom_point(color = 'black', size=3, alpha=0.4)+
  scale_color_manual(values = my_colors)

Now try picking one of the default RColorBrewer themes, using the scale_color_brewer function. See available palettes here

target_countries <- c('Venezuela', 'Vietnam', 'Uruguay')
gap_targ <- gapminder %>% 
  filter(country %in% target_countries)
ggplot(gap_targ, mapping = aes(x = year, lifeExp, color = country)) + 
  geom_line() + 
  geom_point(color = 'black', size=3, alpha=0.4)+
scale_colour_brewer(palette = "Set1")

geom_smooth

Make a plot using geom_smooth to show the average trends of lifeExp over time for each continent. Test out using both linear regression (method = 'lm') and 'loess' models (the default). Also try out turning off the confidence intervals (parameter se in geom_smooth).

For reference, your plots should look something like this:

ggplot(gapminder, aes(year, lifeExp, color = continent)) + 
  geom_smooth(se = FALSE)

ggplot(gapminder, aes(year, lifeExp, color = continent)) + 
  geom_smooth(se = FALSE)

geom_facet

Using only data from up to and including the year 1962, make a faceted scatterplot of gdpPercap vs lifeExp, where there is a separate scatterplot for each year (use facet_wrap()). Make gdpPercap on a log10 scale Add a 'smoothing trend' with geom_smooth() if you want!

It should look like this:

gap_early <- gapminder %>% filter(year <= 1962)
ggplot(gap_early, mapping = aes(gdpPercap, lifeExp)) + 
  scale_x_log10() +
  geom_point() +
  geom_smooth(se = FALSE) +
  facet_wrap(~year)

gap_early <- gapminder %>% filter(year <= 1962)
ggplot(gap_early, mapping = aes(gdpPercap, lifeExp)) + 
  scale_x_log10() +
  geom_point() +
  geom_smooth(se = FALSE) +
  facet_wrap(~year)

Adding text

Make a table called continent_stats which has the average lifeExp and average gdpPercap for each continent. Then use to it make a scatterplot of avg lifeExp vs avg gdp, and label the points using geom_label_repel() (or geom_text_repel() if you prefer)

Your plot should look like this:

continent_stats <- gapminder %>% 
  group_by(continent) %>% 
  summarise(avg_lifeExp = mean(lifeExp),
            avg_gdpPercap = mean(gdpPercap))
ggplot(continent_stats, mapping = aes(avg_lifeExp, avg_gdpPercap, label = continent)) + 
  geom_point() + 
  geom_label_repel()

continent_stats <- gapminder %>% 
  group_by(continent) %>% 
  summarise(avg_lifeExp = mean(lifeExp),
            avg_gdpPercap = mean(gdpPercap))
ggplot(continent_stats, mapping = aes(avg_lifeExp, avg_gdpPercap, label = continent)) + 
  geom_point() + 
  geom_label_repel()

Plotting distributions

Violin plots

Make a violin plot of lifeExp for each continent

ggplot(gapminder, aes(x=continent, y=lifeExp))+
  geom_violin()

Now make each continent's violin a different fill color (play around with using fill vs color to encode the continent)

Try the fill aesthetic

ggplot(gapminder, aes(x=continent, y=lifeExp, fill=continent))+
  geom_violin()

Try the color aesthetic

ggplot(gapminder, aes(x=continent, y=lifeExp, color=continent))+
  geom_violin()

Boxplot plots

Try it as a boxplot (Make a boxplot plot of lifeExp for each continent )

ggplot(gapminder, aes(x=continent, y=lifeExp))+
  geom_boxplot()

Use fct_reorder to reorder the continents by median lifeExp before making your boxplot. The inputs to fct_reorder are a categorical variable, a value variable to order the categories by, and a summary function. So, fct_reorder(country, gdp, max) would reorder the country categorical variable based on max GDP across years.

gapminder <- gapminder %>% mutate(continent = fct_reorder(continent, lifeExp, median))
ggplot(gapminder, aes(x=continent, y=lifeExp))+
  geom_boxplot()

Try making a density plot with geom_density(), separate continents by fill color

ggplot(gapminder, aes(fill=continent, x=lifeExp))+
  geom_density()

What happens when you make the density plots semi-transparent (i.e. setting alpha to ~0.5)? Try the alpha property to change transparency

ggplot(gapminder, aes(fill=continent, x=lifeExp))+
  geom_density(alpha=0.5)

Using theme

Rotate axis

Now let's make a bar plot of the gdpPercap for countries in the Americas in the year 1982. Recall that geom_col will do this. Your plot should initially look like this:

gap_82 <- gapminder %>% 
  filter(year == 1982, continent == 'Americas')

gap_82 <- gapminder %>% 
  filter(year == 1982, continent == 'Americas') 
ggplot(gap_82, aes(x = country, y = gdpPercap)) + 
  geom_col()

ggplot(gap_82, )

Let's try to figure out how to fix the x-axis labels so they are vertical (and not overlapping). See if you can use Google to figure out how to rotate the x-axis labels in ggplot to make them vertical.

ggplot(gap_82, aes(x = country, y = gdpPercap)) + 
  geom_col() +
  ___

ggplot(gap_82, aes(x = country, y = gdpPercap)) + 
  geom_col() +
  theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1))

Reorder countries

Then see if you can reorder the countries from highest to lowest. Here again you can use the fct_reorder function, inside a mutate to order countries by gdpPercap. Note that there's another optional input to fct_reorder called .desc which is FALSE by default

Your plot should look like this

gap_82 <- gap_82 %>% 
  mutate(country = fct_reorder(country, gdpPercap, .desc = TRUE))
ggplot(gap_82, aes(x = country, y = gdpPercap)) + 
  geom_col() +
  theme(axis.text.x = element_text(angle = 90, hjust = 1))

gap_82 <- gap_82 %>% 
  mutate(country = fct_reorder(country, gdpPercap, .desc = TRUE))
ggplot(gap_82, aes(x = country, y = gdpPercap)) + 
  geom_col() +
  theme(axis.text.x = element_text(angle = 90, hjust = 1))

plotly

Now take the code from the last example in 'Scatterplot basics' to make a scatterplot of gdp vs life expectancy from countries in 1992, and let's make it an interactive scatterplot using the package plotly. All you have to do is save your plot to a variable (say my_plot, and then pass it as an input to the function ggplotly). Try it out! Try adding an aesthetic text (say to encode the country) and see what happens. Your plot should look like this:

my_data <- gapminder %>% 
  filter(year == 1992) %>% 
  mutate(gdp = gdpPercap * pop / 1e9) 
my_plot <- ggplot(my_data, aes(gdp, lifeExp, text = country)) + 
  geom_point(aes(color = continent, size = pop)) + 
  scale_x_log10() +
  guides(size = 'none') +
  xlab('Gross Domestic Product (Billions $)') +
  ylab('Life Expectancy at birth (years)')
ggplotly(my_plot)

## Adding the text as an aesthetic
ggplot(my_data, aes(text = country))

my_data <- gapminder %>% 
  filter(year == 1992) %>% 
  mutate(gdp = gdpPercap * pop / 1e9) 
my_plot <- ggplot(my_data, aes(gdp, lifeExp, text = country)) + 
  geom_point(aes(color = continent, size = pop)) + 
  scale_x_log10() +
  guides(size = 'none') +
  xlab('Gross Domestic Product (Billions $)') +
  ylab('Life Expectancy at birth (years)')
ggplotly(my_plot)

Extra practice: GO analysis

For extra practice, try following along with the ggplot tutorial from HBC here

The top 30 enriched GO biological process gene results are loaded in the table bp_oe

load(here::here('data','GOs_for_plotting.RData'))
bp_oe <- bp_oe %>% head(30)

head(bp_oe)

The tutorial will walk through how to make a plot that looks like this:

library(RColorBrewer)
mypalette <- brewer.pal(3, "YlOrRd")

ggplot(bp_oe) +
  geom_point(aes(x = gene_ratio, y = GO_term, color = -log10(p.value)), 
             size = 2) +
  theme_bw() +
  theme(axis.text.x = element_text(size=rel(1.15)),
        axis.text.y = element_text(size=rel(0.7)),
        axis.title = element_text(size=rel(1.15))) +
  xlab("Gene ratios") +
  ylab("Top 30 significant GO terms") +
  ggtitle("Dotplot of top 30 significant GO terms") +
  theme(plot.title = element_text(hjust=0.5, 
    face = "bold")) +
  scale_color_gradientn(name = "Significance \n (-log10(padj))", colors = mypalette) +
  theme(legend.title = element_text(size=rel(1.15),
    hjust=0.5, 
    face="bold"))