Lab 13: ggplot2 — Improving Graphs & Saving Plots

BI 255 · Bethune-Cookman University · Fall 2026

Author

Dr. Rosie Stanbrook-Buyer

Published

September 23, 2026

Publication-Quality Graphics with ggplot2

The base R graphics system we used in Lab 12 is powerful but requires a lot of manual customisation. The ggplot2 package takes a fundamentally different approach — it implements a grammar of graphics, a consistent set of rules for how plots are built from components. Once you understand the grammar, you can build almost any plot by combining the same small set of building blocks.

ggplot2 is the standard for data visualisation in ecology, genomics, clinical research, and most areas of biological data science. Nearly every figure you see in modern biology journals was made with it or something equivalent.

Learning Objectives

By the end of this lab you will be able to:

Explain the grammar of graphics: data, aesthetics, and geoms
Build geom_histogram(), geom_bar(), geom_boxplot(), and geom_point() plots
Map variables to colour, shape, and size with aes()
Apply and customise themes
Use facet_wrap() to create small-multiple plots
Save publication-quality figures with ggsave()

Part 1: The Grammar of Graphics

Every ggplot2 figure is built from three required components:

The Three Layers of Every ggplot

1. Data — the data frame you are plotting ggplot(data = my_data, ...)

2. Aesthetics — which columns map to which visual properties aes(x = col1, y = col2, colour = col3, ...)

3. Geom — the geometric shape used to represent the data geom_point(), geom_boxplot(), geom_histogram(), etc.

A minimal plot: ggplot(data, aes(x = col)) + geom_histogram()

Additional layers (themes, scales, labels, facets) are added with +.

library(ggplot2)

Dataset: Gene Expression in Two Cell Lines

We will work with a simulated RNA-seq-style dataset comparing gene expression levels across two cell lines (wildtype and a knockout) under two treatment conditions:

set.seed(303)
n_per_group <- 10

gene_expr <- data.frame(
  gene        = paste0("Gene_", sprintf("%02d", rep(1:n_per_group, 6))),
  cell_line   = rep(c("Wildtype", "Knockout"), each = 3 * n_per_group),
  treatment   = rep(rep(c("Control", "Drug_A", "Drug_B"), each = n_per_group), times = 2),
  expression  = c(
    rnorm(n_per_group, mean = 8.2,  sd = 1.8),   # Wildtype control
    rnorm(n_per_group, mean = 11.4, sd = 2.1),   # Wildtype Drug A
    rnorm(n_per_group, mean = 7.9,  sd = 1.6),   # Wildtype Drug B
    rnorm(n_per_group, mean = 5.8,  sd = 1.5),   # Knockout control
    rnorm(n_per_group, mean = 6.3,  sd = 1.7),   # Knockout Drug A
    rnorm(n_per_group, mean = 5.6,  sd = 1.4)    # Knockout Drug B
  ),
  log2_fc     = c(
    rnorm(n_per_group, mean =  0.2, sd = 0.8),
    rnorm(n_per_group, mean =  1.8, sd = 0.9),
    rnorm(n_per_group, mean = -0.3, sd = 0.7),
    rnorm(n_per_group, mean = -1.1, sd = 0.8),
    rnorm(n_per_group, mean = -0.8, sd = 0.9),
    rnorm(n_per_group, mean = -1.3, sd = 0.7)
  )
)

str(gene_expr)

'data.frame':   60 obs. of  5 variables:
 $ gene      : chr  "Gene_01" "Gene_02" "Gene_03" "Gene_04" ...
 $ cell_line : chr  "Wildtype" "Wildtype" "Wildtype" "Wildtype" ...
 $ treatment : chr  "Control" "Control" "Control" "Control" ...
 $ expression: num  7.31 8.7 8.89 8.32 7.16 ...
 $ log2_fc   : num  -0.0792 -1.1112 0.1429 0.6093 -0.6354 ...

Part 2: Histograms with `geom_histogram()`

# Basic histogram
ggplot(gene_expr, aes(x = expression)) +
  geom_histogram(bins = 15, fill = "#2C5F8A", colour = "white") +
  labs(title = "Distribution of Gene Expression Values",
       x     = "Expression Level (log2 CPM)",
       y     = "Count")

Colour by Group

# Separate histograms by cell line, overlaid
ggplot(gene_expr, aes(x = expression, fill = cell_line)) +
  geom_histogram(bins = 12, alpha = 0.6, position = "identity",
                 colour = "white") +
  scale_fill_manual(values = c("Wildtype" = "#2C5F8A",
                               "Knockout" = "#C8102E")) +
  labs(title = "Expression Distribution by Cell Line",
       x     = "Expression Level (log2 CPM)",
       y     = "Count",
       fill  = "Cell Line") +
  theme_classic()

Key aes() Mappings

Mapping	What it controls
`x`	Position on x-axis
`y`	Position on y-axis
`colour`	Outline/line colour
`fill`	Fill colour (inside bars, boxes, etc.)
`shape`	Point shape (for `geom_point`)
`size`	Point or line size
`alpha`	Transparency (0 = invisible, 1 = opaque)

These can be mapped to a variable inside aes(), or set to a fixed value outside aes().

Part 3: Bar Charts with `geom_bar()` and `geom_col()`

# Mean expression per treatment — first summarise the data
library(dplyr)

summary_expr <- gene_expr |>
  group_by(treatment, cell_line) |>
  summarise(mean_expr = mean(expression),
            se_expr   = sd(expression) / sqrt(n()),
            .groups   = "drop")

summary_expr

# A tibble: 6 × 4
  treatment cell_line mean_expr se_expr
  <chr>     <chr>         <dbl>   <dbl>
1 Control   Knockout       5.52   0.360
2 Control   Wildtype       8.08   0.426
3 Drug_A    Knockout       6.39   0.660
4 Drug_A    Wildtype      11.9    0.534
5 Drug_B    Knockout       4.97   0.561
6 Drug_B    Wildtype       8.20   0.512

# Grouped bar chart using geom_col() (for pre-computed values)
ggplot(summary_expr, aes(x = treatment, y = mean_expr, fill = cell_line)) +
  geom_col(position = "dodge", width = 0.7) +
  geom_errorbar(aes(ymin = mean_expr - se_expr,
                    ymax = mean_expr + se_expr),
                position = position_dodge(0.7),
                width    = 0.2) +
  scale_fill_manual(values = c("Wildtype" = "#2C5F8A",
                               "Knockout" = "#C8102E")) +
  labs(title = "Mean Gene Expression by Treatment and Cell Line",
       x     = "Treatment",
       y     = "Mean Expression (log2 CPM) ± SE",
       fill  = "Cell Line") +
  theme_classic(base_size = 13)

geom_bar() vs geom_col()

geom_bar() counts rows and plots the count — use it when your data is unaggregated.

geom_col() uses values already in your data frame — use it when you have pre-computed means or totals.

For scientific figures, you almost always want geom_col() with a summary data frame.

Part 4: Box Plots with `geom_boxplot()`

# Box plot comparing expression across treatments, split by cell line
ggplot(gene_expr, aes(x = treatment, y = expression, fill = cell_line)) +
  geom_boxplot(alpha = 0.7, outlier.shape = NA) +
  geom_jitter(aes(colour = cell_line),
              position = position_jitterdodge(jitter.width = 0.15),
              size     = 1.5, alpha = 0.6) +
  scale_fill_manual(values = c("Wildtype"  = "#2C5F8A",
                               "Knockout"  = "#C8102E")) +
  scale_colour_manual(values = c("Wildtype" = "#1A3D5C",
                                 "Knockout" = "#8B0000")) +
  labs(title  = "Gene Expression Distribution by Treatment",
       x      = "Treatment",
       y      = "Expression Level (log2 CPM)",
       fill   = "Cell Line",
       colour = "Cell Line") +
  theme_classic(base_size = 13) +
  theme(legend.position = "top")

Part 5: Scatter Plots with `geom_point()`

# Scatter plot: expression level vs log2 fold change
ggplot(gene_expr, aes(x = expression, y = log2_fc,
                      colour = cell_line, shape = treatment)) +
  geom_point(size = 2.5, alpha = 0.8) +
  geom_hline(yintercept = 0, linetype = "dashed", colour = "grey50") +
  scale_colour_manual(values = c("Wildtype" = "#2C5F8A",
                                 "Knockout" = "#C8102E")) +
  labs(title  = "Expression Level vs Log2 Fold Change",
       x      = "Baseline Expression (log2 CPM)",
       y      = "Log2 Fold Change vs Control",
       colour = "Cell Line",
       shape  = "Treatment") +
  theme_bw(base_size = 13)

Why Log2 Fold Change?

In genomics, fold change (FC) tells you how much a gene’s expression changed between conditions. Log2 FC is the standard because: - Log2 FC = 1 means expression doubled - Log2 FC = -1 means expression halved - Log2 FC = 0 means no change

This symmetric scale makes gains and losses directly comparable.

Part 6: Faceting — Small Multiples

Faceting creates separate panels for each level of a variable. This is one of ggplot2’s most powerful features.

# One panel per treatment
ggplot(gene_expr, aes(x = expression, fill = cell_line)) +
  geom_histogram(bins = 10, alpha = 0.7, position = "identity",
                 colour = "white") +
  facet_wrap(~ treatment, ncol = 3) +
  scale_fill_manual(values = c("Wildtype" = "#2C5F8A",
                               "Knockout" = "#C8102E")) +
  labs(title = "Expression Distribution by Treatment",
       x     = "Expression (log2 CPM)",
       y     = "Count",
       fill  = "Cell Line") +
  theme_bw(base_size = 12) +
  theme(strip.background = element_rect(fill = "#E8EEF4"))

Part 7: Themes and Saving Plots

Themes

# The same plot with different themes
p <- ggplot(gene_expr, aes(x = treatment, y = expression, fill = cell_line)) +
  geom_boxplot(alpha = 0.7) +
  scale_fill_manual(values = c("Wildtype" = "#2C5F8A", "Knockout" = "#C8102E")) +
  labs(x = "Treatment", y = "Expression (log2 CPM)", fill = "Cell Line")

p + theme_classic()    # Clean, publication-style

Common ggplot2 Themes

Theme	Style
`theme_classic()`	Clean axes, no grid — good for papers
`theme_bw()`	White background with grid lines
`theme_minimal()`	Minimal background, subtle grid
`theme_void()`	Empty — useful for maps and diagrams

theme_classic() or theme_bw() are the most commonly required for journal submission.

Saving with `ggsave()`

# Build a final figure
final_plot <- ggplot(gene_expr, aes(x = treatment, y = expression,
                                     fill = cell_line)) +
  geom_boxplot(alpha = 0.7, outlier.shape = NA) +
  geom_jitter(aes(colour = cell_line),
              position = position_jitterdodge(0.15),
              size = 1.2, alpha = 0.5) +
  scale_fill_manual(values = c("Wildtype" = "#2C5F8A",
                               "Knockout" = "#C8102E")) +
  scale_colour_manual(values = c("Wildtype" = "#1A3D5C",
                                 "Knockout" = "#8B0000")) +
  labs(title  = "Gene Expression Across Treatments",
       x      = "Treatment",
       y      = "Expression Level (log2 CPM)",
       fill   = "Cell Line",
       colour = "Cell Line") +
  theme_classic(base_size = 13) +
  theme(legend.position = "top",
        plot.title      = element_text(face = "bold"))

# Save to file
ggsave(filename = "gene_expression_boxplot.png",
       plot     = final_plot,
       width    = 7,
       height   = 5,
       dpi      = 300)

cat("Plot saved. Check your Files panel in Posit Cloud.\n")

Plot saved. Check your Files panel in Posit Cloud.

ggsave() Arguments

Argument	What it does
`filename`	Output file name; extension sets format (.png, .pdf, .svg)
`plot`	The ggplot object to save (defaults to last plot shown)
`width`, `height`	Figure dimensions in inches
`dpi`	Resolution; 300 dpi is standard for print publication

Always use 300 dpi when submitting to a journal or printing a poster.

3-Minute Knowledge Check

Close your notes. Answer these on your own — you have 3 minutes. We’ll go through the answers together after.

Knowledge Check Questions

1. What are the three required components of every ggplot2 figure?

2. In aes(x = gene, y = expression, colour = treatment), the colour = treatment mapping means:

a) All points will be coloured the same    b) Points will be coloured according to the values in the `treatment` column    c) The word "treatment" will appear in the colour legend    d) The colour is set to the value stored in the variable `treatment`

3. You want separate histogram panels for each level of a variable called tissue_type. Which ggplot2 function creates these small multiples?

4. What is the difference between geom_bar() and geom_col()?

5. True or False: ggsave("figure.pdf", dpi = 300) will save the plot as a PDF at 300 dpi resolution.

Answers (Only reveal if you have answered the above questions)

1. Data, aesthetics (aes()), and a geom (e.g., geom_point()). Data tells ggplot where to look; aesthetics map columns to visual properties; the geom determines the geometric shape used to represent the data.

2. b) Points will be coloured according to the values in the treatment column — ggplot2 automatically assigns a different colour to each unique value in treatment and creates a legend. To set all points to one colour regardless of the data, you would write colour = "blue" outside the aes() call.

3. facet_wrap(~ tissue_type) — this creates one panel per level of tissue_type. You can control the number of columns with ncol and allow axis scales to vary with scales = "free".

4. geom_bar() counts the number of rows for each x-value and plots that count. geom_col() uses a y-variable you provide directly (pre-computed values like means). For scientific figures with calculated summaries, always use geom_col().

5. False (partially) — it will save as a PDF, but dpi is only relevant for raster formats (PNG, JPEG). PDFs are vector graphics and scale infinitely regardless of dpi. For print publication, PDF or SVG are actually preferable because they remain sharp at any size.

Lab 13 Checklist

Before you leave, make sure you can:

Load ggplot2 with library(ggplot2)
Build a ggplot figure by combining ggplot(), aes(), and a geom
Create geom_histogram(), geom_col(), geom_boxplot(), and geom_point() plots
Map variables to colour, fill, shape, and size inside aes()
Set fixed aesthetic values (e.g., fill = "blue") outside aes()
Add error bars with geom_errorbar()
Apply a theme with theme_classic() or theme_bw()
Create faceted plots with facet_wrap()
Save a plot to file with ggsave() at 300 dpi

Bonus Challenge

Using the gene_expr data frame, create a volcano-plot-style scatter plot with expression on the x-axis and log2_fc on the y-axis. Colour points red if |log2_fc| > 1 and blue otherwise. Add a horizontal dashed line at y = 0 and vertical dashed lines at x = ±1. Apply theme_classic(). Save it as a 6 × 5 inch PNG.

Before Next Class (Friday)

Friday (Lab 14): Contingency tables, mosaic plots, odds ratios, and chi-square tests
Read R for Data Science Ch. 7–8 (if not done)
Have ggplot2 installed and working before class

--- title: "Lab 13: ggplot2 — Improving Graphs & Saving Plots" subtitle: "BI 255 · Bethune-Cookman University · Fall 2026" author: "Dr. Rosie Stanbrook-Buyer" date: "September 23, 2026" format: html: theme: cosmo toc: true toc-location: left toc-title: "In This Lab" toc-depth: 3 number-sections: false code-fold: false code-tools: true highlight-style: github smooth-scroll: true embed-resources: true callout-appearance: default execute: warning: false message: false echo: true --- ```{r setup, include=FALSE} knitr::opts_chunk$set(echo = TRUE) ``` ------------------------------------------------------------------------ ## Publication-Quality Graphics with ggplot2 The base R graphics system we used in Lab 12 is powerful but requires a lot of manual customisation. The `ggplot2` package takes a fundamentally different approach — it implements a **grammar of graphics**, a consistent set of rules for how plots are built from components. Once you understand the grammar, you can build almost any plot by combining the same small set of building blocks. `ggplot2` is the standard for data visualisation in ecology, genomics, clinical research, and most areas of biological data science. Nearly every figure you see in modern biology journals was made with it or something equivalent. ::: callout-note ## Learning Objectives By the end of this lab you will be able to: 1. Explain the grammar of graphics: data, aesthetics, and geoms 2. Build `geom_histogram()`, `geom_bar()`, `geom_boxplot()`, and `geom_point()` plots 3. Map variables to colour, shape, and size with `aes()` 4. Apply and customise themes 5. Use `facet_wrap()` to create small-multiple plots 6. Save publication-quality figures with `ggsave()` ::: ------------------------------------------------------------------------ ## Part 1: The Grammar of Graphics Every ggplot2 figure is built from three required components: ::: callout-important ## The Three Layers of Every ggplot **1. Data** — the data frame you are plotting `ggplot(data = my_data, ...)` **2. Aesthetics** — which columns map to which visual properties `aes(x = col1, y = col2, colour = col3, ...)` **3. Geom** — the geometric shape used to represent the data `geom_point()`, `geom_boxplot()`, `geom_histogram()`, etc. A minimal plot: `ggplot(data, aes(x = col)) + geom_histogram()` Additional layers (themes, scales, labels, facets) are added with `+`. ::: ```{r} library(ggplot2) ``` ### Dataset: Gene Expression in Two Cell Lines We will work with a simulated RNA-seq-style dataset comparing gene expression levels across two cell lines (wildtype and a knockout) under two treatment conditions: ```{r} set.seed(303) n_per_group <- 10 gene_expr <- data.frame( gene = paste0("Gene_", sprintf("%02d", rep(1:n_per_group, 6))), cell_line = rep(c("Wildtype", "Knockout"), each = 3 * n_per_group), treatment = rep(rep(c("Control", "Drug_A", "Drug_B"), each = n_per_group), times = 2), expression = c( rnorm(n_per_group, mean = 8.2, sd = 1.8), # Wildtype control rnorm(n_per_group, mean = 11.4, sd = 2.1), # Wildtype Drug A rnorm(n_per_group, mean = 7.9, sd = 1.6), # Wildtype Drug B rnorm(n_per_group, mean = 5.8, sd = 1.5), # Knockout control rnorm(n_per_group, mean = 6.3, sd = 1.7), # Knockout Drug A rnorm(n_per_group, mean = 5.6, sd = 1.4) # Knockout Drug B ), log2_fc = c( rnorm(n_per_group, mean = 0.2, sd = 0.8), rnorm(n_per_group, mean = 1.8, sd = 0.9), rnorm(n_per_group, mean = -0.3, sd = 0.7), rnorm(n_per_group, mean = -1.1, sd = 0.8), rnorm(n_per_group, mean = -0.8, sd = 0.9), rnorm(n_per_group, mean = -1.3, sd = 0.7) ) ) str(gene_expr) ``` ------------------------------------------------------------------------ ## Part 2: Histograms with `geom_histogram()` ```{r} # Basic histogram ggplot(gene_expr, aes(x = expression)) + geom_histogram(bins = 15, fill = "#2C5F8A", colour = "white") + labs(title = "Distribution of Gene Expression Values", x = "Expression Level (log2 CPM)", y = "Count") ``` ### Colour by Group ```{r} # Separate histograms by cell line, overlaid ggplot(gene_expr, aes(x = expression, fill = cell_line)) + geom_histogram(bins = 12, alpha = 0.6, position = "identity", colour = "white") + scale_fill_manual(values = c("Wildtype" = "#2C5F8A", "Knockout" = "#C8102E")) + labs(title = "Expression Distribution by Cell Line", x = "Expression Level (log2 CPM)", y = "Count", fill = "Cell Line") + theme_classic() ``` ::: callout-note ## Key `aes()` Mappings | Mapping | What it controls | |---------|-----------------| | `x` | Position on x-axis | | `y` | Position on y-axis | | `colour` | Outline/line colour | | `fill` | Fill colour (inside bars, boxes, etc.) | | `shape` | Point shape (for `geom_point`) | | `size` | Point or line size | | `alpha` | Transparency (0 = invisible, 1 = opaque) | These can be **mapped to a variable** inside `aes()`, or **set to a fixed value** outside `aes()`. ::: ------------------------------------------------------------------------ ## Part 3: Bar Charts with `geom_bar()` and `geom_col()` ```{r} # Mean expression per treatment — first summarise the data library(dplyr) summary_expr <- gene_expr |> group_by(treatment, cell_line) |> summarise(mean_expr = mean(expression), se_expr = sd(expression) / sqrt(n()), .groups = "drop") summary_expr ``` ```{r} # Grouped bar chart using geom_col() (for pre-computed values) ggplot(summary_expr, aes(x = treatment, y = mean_expr, fill = cell_line)) + geom_col(position = "dodge", width = 0.7) + geom_errorbar(aes(ymin = mean_expr - se_expr, ymax = mean_expr + se_expr), position = position_dodge(0.7), width = 0.2) + scale_fill_manual(values = c("Wildtype" = "#2C5F8A", "Knockout" = "#C8102E")) + labs(title = "Mean Gene Expression by Treatment and Cell Line", x = "Treatment", y = "Mean Expression (log2 CPM) ± SE", fill = "Cell Line") + theme_classic(base_size = 13) ``` ::: callout-tip ## `geom_bar()` vs `geom_col()` `geom_bar()` counts rows and plots the count — use it when your data is unaggregated. `geom_col()` uses values already in your data frame — use it when you have pre-computed means or totals. For scientific figures, you almost always want `geom_col()` with a summary data frame. ::: ------------------------------------------------------------------------ ## Part 4: Box Plots with `geom_boxplot()` ```{r} # Box plot comparing expression across treatments, split by cell line ggplot(gene_expr, aes(x = treatment, y = expression, fill = cell_line)) + geom_boxplot(alpha = 0.7, outlier.shape = NA) + geom_jitter(aes(colour = cell_line), position = position_jitterdodge(jitter.width = 0.15), size = 1.5, alpha = 0.6) + scale_fill_manual(values = c("Wildtype" = "#2C5F8A", "Knockout" = "#C8102E")) + scale_colour_manual(values = c("Wildtype" = "#1A3D5C", "Knockout" = "#8B0000")) + labs(title = "Gene Expression Distribution by Treatment", x = "Treatment", y = "Expression Level (log2 CPM)", fill = "Cell Line", colour = "Cell Line") + theme_classic(base_size = 13) + theme(legend.position = "top") ``` ------------------------------------------------------------------------ ## Part 5: Scatter Plots with `geom_point()` ```{r} # Scatter plot: expression level vs log2 fold change ggplot(gene_expr, aes(x = expression, y = log2_fc, colour = cell_line, shape = treatment)) + geom_point(size = 2.5, alpha = 0.8) + geom_hline(yintercept = 0, linetype = "dashed", colour = "grey50") + scale_colour_manual(values = c("Wildtype" = "#2C5F8A", "Knockout" = "#C8102E")) + labs(title = "Expression Level vs Log2 Fold Change", x = "Baseline Expression (log2 CPM)", y = "Log2 Fold Change vs Control", colour = "Cell Line", shape = "Treatment") + theme_bw(base_size = 13) ``` ::: callout-note ## Why Log2 Fold Change? In genomics, fold change (FC) tells you how much a gene's expression changed between conditions. Log2 FC is the standard because: - Log2 FC = 1 means expression **doubled** - Log2 FC = -1 means expression **halved** - Log2 FC = 0 means **no change** This symmetric scale makes gains and losses directly comparable. ::: ------------------------------------------------------------------------ ## Part 6: Faceting — Small Multiples Faceting creates separate panels for each level of a variable. This is one of ggplot2's most powerful features. ```{r} # One panel per treatment ggplot(gene_expr, aes(x = expression, fill = cell_line)) + geom_histogram(bins = 10, alpha = 0.7, position = "identity", colour = "white") + facet_wrap(~ treatment, ncol = 3) + scale_fill_manual(values = c("Wildtype" = "#2C5F8A", "Knockout" = "#C8102E")) + labs(title = "Expression Distribution by Treatment", x = "Expression (log2 CPM)", y = "Count", fill = "Cell Line") + theme_bw(base_size = 12) + theme(strip.background = element_rect(fill = "#E8EEF4")) ``` ------------------------------------------------------------------------ ## Part 7: Themes and Saving Plots ### Themes ```{r} # The same plot with different themes p <- ggplot(gene_expr, aes(x = treatment, y = expression, fill = cell_line)) + geom_boxplot(alpha = 0.7) + scale_fill_manual(values = c("Wildtype" = "#2C5F8A", "Knockout" = "#C8102E")) + labs(x = "Treatment", y = "Expression (log2 CPM)", fill = "Cell Line") p + theme_classic() # Clean, publication-style ``` ::: callout-tip ## Common ggplot2 Themes | Theme | Style | |-------|-------| | `theme_classic()` | Clean axes, no grid — good for papers | | `theme_bw()` | White background with grid lines | | `theme_minimal()` | Minimal background, subtle grid | | `theme_void()` | Empty — useful for maps and diagrams | `theme_classic()` or `theme_bw()` are the most commonly required for journal submission. ::: ### Saving with `ggsave()` ```{r} # Build a final figure final_plot <- ggplot(gene_expr, aes(x = treatment, y = expression, fill = cell_line)) + geom_boxplot(alpha = 0.7, outlier.shape = NA) + geom_jitter(aes(colour = cell_line), position = position_jitterdodge(0.15), size = 1.2, alpha = 0.5) + scale_fill_manual(values = c("Wildtype" = "#2C5F8A", "Knockout" = "#C8102E")) + scale_colour_manual(values = c("Wildtype" = "#1A3D5C", "Knockout" = "#8B0000")) + labs(title = "Gene Expression Across Treatments", x = "Treatment", y = "Expression Level (log2 CPM)", fill = "Cell Line", colour = "Cell Line") + theme_classic(base_size = 13) + theme(legend.position = "top", plot.title = element_text(face = "bold")) # Save to file ggsave(filename = "gene_expression_boxplot.png", plot = final_plot, width = 7, height = 5, dpi = 300) cat("Plot saved. Check your Files panel in Posit Cloud.\n") ``` ::: callout-important ## `ggsave()` Arguments | Argument | What it does | |----------|-------------| | `filename` | Output file name; extension sets format (.png, .pdf, .svg) | | `plot` | The ggplot object to save (defaults to last plot shown) | | `width`, `height` | Figure dimensions in inches | | `dpi` | Resolution; 300 dpi is standard for print publication | Always use 300 dpi when submitting to a journal or printing a poster. ::: ------------------------------------------------------------------------ ## 3-Minute Knowledge Check *Close your notes. Answer these on your own — you have 3 minutes. We'll go through the answers together after.* ::: callout-caution ## Knowledge Check Questions **1.** What are the three required components of every ggplot2 figure? **2.** In `aes(x = gene, y = expression, colour = treatment)`, the `colour = treatment` mapping means: a) All points will be coloured the same b) Points will be coloured according to the values in the `treatment` column c) The word "treatment" will appear in the colour legend d) The colour is set to the value stored in the variable `treatment` **3.** You want separate histogram panels for each level of a variable called `tissue_type`. Which ggplot2 function creates these small multiples? **4.** What is the difference between `geom_bar()` and `geom_col()`? **5.** True or False: `ggsave("figure.pdf", dpi = 300)` will save the plot as a PDF at 300 dpi resolution. ::: ::: {.callout-caution collapse="true"} ## Answers (Only reveal if you have answered the above questions) **1.** **Data**, **aesthetics** (`aes()`), and a **geom** (e.g., `geom_point()`). Data tells ggplot where to look; aesthetics map columns to visual properties; the geom determines the geometric shape used to represent the data. **2.** b) Points will be coloured according to the values in the `treatment` column — ggplot2 automatically assigns a different colour to each unique value in `treatment` and creates a legend. To set all points to one colour regardless of the data, you would write `colour = "blue"` **outside** the `aes()` call. **3.** `facet_wrap(~ tissue_type)` — this creates one panel per level of `tissue_type`. You can control the number of columns with `ncol` and allow axis scales to vary with `scales = "free"`. **4.** `geom_bar()` counts the number of rows for each x-value and plots that count. `geom_col()` uses a y-variable you provide directly (pre-computed values like means). For scientific figures with calculated summaries, always use `geom_col()`. **5.** **False (partially)** — it will save as a PDF, but `dpi` is only relevant for raster formats (PNG, JPEG). PDFs are vector graphics and scale infinitely regardless of dpi. For print publication, PDF or SVG are actually preferable because they remain sharp at any size. ::: ------------------------------------------------------------------------ ## Lab 13 Checklist Before you leave, make sure you can: - [ ] Load ggplot2 with `library(ggplot2)` - [ ] Build a ggplot figure by combining `ggplot()`, `aes()`, and a geom - [ ] Create `geom_histogram()`, `geom_col()`, `geom_boxplot()`, and `geom_point()` plots - [ ] Map variables to colour, fill, shape, and size inside `aes()` - [ ] Set fixed aesthetic values (e.g., `fill = "blue"`) outside `aes()` - [ ] Add error bars with `geom_errorbar()` - [ ] Apply a theme with `theme_classic()` or `theme_bw()` - [ ] Create faceted plots with `facet_wrap()` - [ ] Save a plot to file with `ggsave()` at 300 dpi ::: callout-tip ## Bonus Challenge Using the `gene_expr` data frame, create a volcano-plot-style scatter plot with `expression` on the x-axis and `log2_fc` on the y-axis. Colour points red if `|log2_fc| > 1` and blue otherwise. Add a horizontal dashed line at y = 0 and vertical dashed lines at x = ±1. Apply `theme_classic()`. Save it as a 6 × 5 inch PNG. ::: ------------------------------------------------------------------------ ## Before Next Class (Friday) - Friday (Lab 14): Contingency tables, mosaic plots, odds ratios, and chi-square tests - Read **R for Data Science Ch. 7–8** (if not done) - Have ggplot2 installed and working before class ------------------------------------------------------------------------