Graphs from tables

Imagine an experiment:

• treatment and control groups
• protein concentration as time-series

Graphs from tables

And you want to plot it as two lines using Excel

Graphs from tables

R gives you a professional looking figure

Read a table in R

If this table is in dummy.csv:

[bash]$ cat dummy.csv
sample_id,condition,time,concentration
S1,control,1,5.1
S2,control,2,4.8
S3,control,3,5.3
S4,control,4,5.0
S5,control,5,5.2
S6,treated,1,6.7
S7,treated,2,6.9
S8,treated,3,7.0
S9,treated,4,6.8
S10,treated,5,6.6
[bash]$

Here’s how we can read this table into an R object.

df = read.csv("dummy.csv")

Peek into tables

Let us look at the contents of this table

df

##    sample_id condition time concentration
## 1         S1   control    1           5.1
## 2         S2   control    2           4.8
## 3         S3   control    3           5.3
## 4         S4   control    4           5.0
## 5         S5   control    5           5.2
## 6         S6   treated    1           6.7
## 7         S7   treated    2           6.9
## 8         S8   treated    3           7.0
## 9         S9   treated    4           6.8
## 10       S10   treated    5           6.6

Peek into tables

We can also get useful information about it.

# Dimensions: rows and columns
dim(df)

## [1] 10  4

# Structure: variable names and types
str(df)

## 'data.frame':    10 obs. of  4 variables:
##  $ sample_id    : chr  "S1" "S2" "S3" "S4" ...
##  $ condition    : chr  "control" "control" "control" "control" ...
##  $ time         : int  1 2 3 4 5 1 2 3 4 5
##  $ concentration: num  5.1 4.8 5.3 5 5.2 6.7 6.9 7 6.8 6.6

Peek into tables

We can also get useful information about it.

# Summary stats
summary(df)

##   sample_id          condition              time   concentration  
##  Length:10          Length:10          Min.   :1   Min.   :4.800  
##  Class :character   Class :character   1st Qu.:2   1st Qu.:5.125  
##  Mode  :character   Mode  :character   Median :3   Median :5.950  
##                                        Mean   :3   Mean   :5.940  
##                                        3rd Qu.:4   3rd Qu.:6.775  
##                                        Max.   :5   Max.   :7.000

Peek into tables

Or summarize values in columns.

# Count values per group
table(df$condition)

## 
## control treated 
##       5       5

# Count values per group
table(df$time)

## 
## 1 2 3 4 5 
## 2 2 2 2 2

From tables to plots in R

Plotting in R is not so different from Excel

• Excel and ggplot2 speak the same language
• Map columns to axis: time is x-axis and concentration is y-axis

Line graph

ggplot(data = df) +
  aes(x = time, y = concentration, group = condition, color = condition) +
  geom_line()

How about just points?

ggplot(data = df) +
  aes(x = time, y = concentration, group = condition, color = condition) +
  geom_point()

How about lines AND points?

ggplot(data = df) +
  aes(x = time, y = concentration, group = condition, color = condition) +
  geom_line() + geom_point()

Can we start x-axis from 0?

ggplot(data = df) +
  aes(x = time, y = concentration, group = condition, color = condition) +
  geom_line() + geom_point() + ylim(c(0,10))

Can we fix the weird x-axis tics?

ggplot(data = df) +
  aes(x = time, y = concentration, group = condition, color = condition) +
  geom_line() + geom_point() + scale_y_continuous(breaks=seq(0,10,1), limits=c(0,10))

What about that professional look?

ggplot(data = df) +
  aes(x = time, y = concentration, group = condition, color = condition) +
  geom_line() + geom_point() + scale_y_continuous(breaks=seq(0,10,1), limits=c(0,10)) +
  theme_classic()

🎨 The basics of ggplot2

Anatomy of the ggplot function

• ggplot2 builds plots layer by layer.
  
• You always start with:

ggplot(data = <data>, aes(x = <x-column>, y = <y-column>))

or

ggplot(data = <data>) + aes(x = <x-column>, y = <y-column>)

Then, you add a geometry layer: geom_col(), geom_point(), etc.

• geom_point(): for points
• geom_lines(): for lines
• geom_bar(): for barplot
• geom_boxplot(): for boxplot

Recap: Key components of the ggplot function

• ggplot() sets up the plot object and links data to axes.
• aes() maps variables to visual properties (x-axis, y-axis, colors, etc.).
• geom_*() plots data based on y values.
• You can customize it with themes, labels, and colors.
• More in the hands-on sessions.

Why use ggplot?

Why Join Tables?

Biologists often have different pieces of information in separate tables:

• One table has sample metadata (e.g., treatment).
• Another has experimental results (e.g., protein levels).

We want to combine them using a common column (like Sample ID).

Connect protein concentrations to samples

left_join(sample_metadata, protein_conc, by="sample_id")

• Each row now has both sample metadata and result.
• We used a common key (sample_id) to match.

Connect protein concentrations to samples to animals

left_join(left_join(sample_metadata, animal_metadata, by="animal_id"), protein_conc, by="sample_id")

🔗 The pipe operator %>%

What is the pipe operator?

• denoted by '%>%'
• comes from magrittr package.
• chains operations in a readable, step-by-step way.

E.g.

df %>% something() %>% somethingelse()

Think: “and then”.

Remember our left_join()?

Without the pipe

left_join(left_join(sample_metadata, animal_metadata, by="animal_id"), protein_conc, by="sample_id")

With the pipe

sample_metadata %>%
  left_join(animal_metadata, by="animal_id") %>%
  left_join(protein_conc, by="sample_id")

• Easier to read: do this, then this, then this.
• Reduces nested parentheses and clutter.

Thanks! Any Questions?