Working with Packages

What are packages?

R comes with a set of built-in functions. Packages extend R with thousands of additional functions written by other developers. Think of packages as toolboxes you can add to your workshop.

There are two main sources for R packages:

• CRAN is the default package repository. It hosts general-purpose packages for statistics, visualization, and data manipulation.
• Bioconductor is a specialized repository for genomics and bioinformatics packages. Most tools you will use for analyzing sequencing data come from Bioconductor.

Installing from CRAN

Use install.packages() to install a package from CRAN. You only need to install a package once.

install.packages("ggplot2")

To install multiple packages at once, pass a vector of package names.

install.packages(c("dplyr", "readr", "tidyr"))

Installing from Bioconductor

Bioconductor is the main source for genomics and bioinformatics packages in R. Packages like DESeq2, GenomicRanges, and Biostrings all live here.

First, install the BiocManager package from CRAN. This is a helper package that manages Bioconductor installations.

install.packages("BiocManager")

Then use BiocManager::install() to install Bioconductor packages.

BiocManager::install("DESeq2")

You can install multiple Bioconductor packages at once the same way.

BiocManager::install(c("GenomicRanges", "Biostrings"))

BiocManager::install() also works for CRAN packages. Some bioinformaticians use it as their single installation command for everything.

Loading packages

Installing a package downloads it to your computer. Loading a package makes its functions available in your current R session. Use library() to load a package.

library(ggplot2)

You must load packages every time you start a new R session. Installation is a one-time step. Loading is a per-session step.

Checking installed packages

You can check whether a specific package is installed on your system.

"ggplot2" %in% rownames(installed.packages())

[1] TRUE

"nonexistent_pkg" %in% rownames(installed.packages())

[1] FALSE

This is useful in scripts where you want to verify dependencies before running an analysis.

Package versions

Check the version of an installed package with packageVersion().

packageVersion("ggplot2")

[1] '4.0.2'

Knowing your package versions is important for reproducibility. Many bioinformatics papers report the exact versions used in their analyses.

Namespace conflicts

Sometimes two packages define functions with the same name. When you load the second package, its function “masks” the one from the first package. R warns you when this happens.

A common example is filter(). The built-in stats package has a filter() function for time series. The dplyr package also has a filter() function for subsetting data frames. Loading dplyr masks the stats version.

library(dplyr)

Attaching package: 'dplyr'

The following objects are masked from 'package:stats':

    filter, lag

The following objects are masked from 'package:base':

    intersect, setdiff, setequal, union

To use a specific version, prefix the function with the package name and ::. This is called the namespace operator.

df <- data.frame(gene = c("TP53", "BRCA1", "EGFR"), pvalue = c(0.001, 0.5, 0.03))
dplyr::filter(df, pvalue < 0.05)

  gene pvalue
1 TP53  0.001
2 EGFR  0.030

Using package::function() syntax removes all ambiguity. It is good practice whenever you know a conflict exists.

Listing loaded packages

To see which packages are currently loaded in your session, use (.packages()).

(.packages())

[1] "dplyr"     "ggplot2"   "stats"     "graphics"  "grDevices" "utils"
[7] "datasets"  "methods"   "base"

The output includes both packages you loaded explicitly and packages that R loads automatically at startup.

Key bioinformatics packages

Here are some of the most widely used packages in bioinformatics. All are available through Bioconductor unless noted otherwise.

Package	Source	Description
DESeq2	Bioconductor	Differential gene expression analysis from RNA-seq count data
edgeR	Bioconductor	Differential expression analysis using negative binomial models
GenomicRanges	Bioconductor	Representation and manipulation of genomic intervals and coordinates
Biostrings	Bioconductor	Efficient handling of DNA, RNA, and protein sequences
AnnotationDbi	Bioconductor	Interface to gene annotation databases
Seurat	CRAN	Analysis of single-cell RNA-seq data
clusterProfiler	Bioconductor	Gene set enrichment and pathway analysis

You will not need all of these at once. Install them as your projects require them.

Reproducibility with renv

As your analyses grow, keeping track of package versions becomes critical. The renv package helps with this. It creates a snapshot of all package versions used in a project. Collaborators can then restore the exact same versions on their machines.

You do not need to learn renv right away. Just know it exists for when you start sharing analyses or publishing results.

Quick reference

Task	Command
Install from CRAN	install.packages("pkg")
Install from Bioconductor	BiocManager::install("pkg")
Load a package	library(pkg)
Check if installed	"pkg" %in% rownames(installed.packages())
Check version	packageVersion("pkg")
Use specific namespace	package::function()
List loaded packages	(.packages())

Next steps

Now that you can install and manage packages, you are ready to learn the most popular collection of R packages for data science. Continue to Tidyverse Essentials.