You do not need to run these codes inside the OrchestraPlatform’s RStudio instance that we are using for the the in-person workshop (although some of them can be). This lesson is to help you later to get everything you would need to use Bioconductor on your own computer!
If you are in this workshop, you likely already have R and RStudio install on your own computer. If not, Bioconductor’s Carpentries-style Introduction to R workshop has detailed instructions for Windows, macOS and Linux. It also has instructions on how to check for updates if you installed R/RStudio at some point in the past but have not updated in over 6 months - R/RStudio/Bioconductor are constantly being improved.
BiocManager from CRAN
The Comprehensive R Archive Network (CRAN) is
the main R project’s repository of over 20,000 packages. This is what is
automatically used whenever you use install.packages() to
install a desired package. Bioconductor maintains it’s own repository of
3500+ packages, and to download packages from there you first need to
down the BioCManager package from CRAN:
install.packages("BiocManager")Next, use the BiocManager::install() function to install
the core Bioconductor packages. The :: is a way to call the
install() function without having to do
library(BiocManager) first, and it is the Bioconductor
convention to call it that way to remind you that you are installing
from Bioconductor. Calling without any argument will check your version
of R and get the most recent Bioconductor version that is compatible
with it. Or you can specify a particular version; Bioconductor’s install page always uses the most current
version.
# This will always get the most current Bioc version that is compatible with your R version:
BiocManager::install()
# This will get a specific Bioc version, which as of Aug 2026 is 3.23:
BiocManager::install(version = "3.23")BiocManager::install() also checks if you need to
update installed packages
You may notice after running BiocManager::install()
output that lists Old packages: and then asks you if you
want to Update all/some/none? [a/s/n]:. This is because a
default argument is update = TRUE. Two important
points:
Never update packages if you are in the middle
of working and already have loaded packages as trying to update a loaded
package can cause problems. Instead, only update in a fresh R session
before doing any library("somepackage").
To turn this off, simply do
BiocManager::install(update = FALSE)
To check what version of Bioconductor you have installed, you can do this (yes, run this in the workshop!)
BiocManager::version()## [1] '3.24'The BiocManager::install() is just a wrapper around
install.packages() telling it to check both CRAN and
Bioconductor for the packages you ask for. This is because many
Bioconductor packages have CRAN package dependencies and when you
install a package, you also need to install any packages it needs to
work. The great thing about this is you can use
BiocManager::install() to install any package and not have
to remember if it from CRAN or Bioconductor. If you know the name of the
packages you want, you can call them singly or in a character
vector:
With over 3500 packages, it can be difficult to find a package that does what you need. To help with this, Bioconductor has the biocViews page to partition packages first by type (Software/AnnotationData/ExperimentData/Workflow) and then the packages are listed below in order of how often they are downloaded. Clicking on the arrow next to any type shows further subdivision and clicking on any one will restrict the package list below. In the “Find biocViews:” search box you can start typing in what you want like “rna” and a popup box showing the defined biocViews terms that match it. In the packages section you can also use the “Search table:” to find packages that match in the package name, Maintainer or Title columns. Try putting in DESeq2.
One you click on a package in the table, it will take you to that package’s page. At the top is some summary information - mouse over any one to get an explanation. These are here to help you figure out if the package is commonly used, actively maintained and answers questions on the support site. There is also lots of other information about the package, of which the two most practical pieces are:
A vignette shows how to use the package to run a typical analysis from start to finish. This is much more useful than the reference manual, often with a runnable analysis with internal data. This allows a user to run through working code, and later modify it to fit their own data. Many package authors use their vignettes to run package demos at various conferences, and videos of many of these are available on Bioconductor’s YouTube channel.
One of the 4 main types of packages on the biocViews, the Workflow packages maybe the most useful when trying to learn how to analyze a particular type of data. Most end-to-end analyses will use many different packages, and the workflow packages walk through all the typical steps and important asides. They are good for self-study as they also contain runnable code with example data, like vignettes, that you can modify for your own data.
Many online books are available as reference volumes on topics such as single-cell data, spatial data, Hi-C data, and microbiome data. They cover broader scope and concept discussion than the practical workflows. They are regularly built so that the code within them is executable instead of getting outdated like static teaching materials.
There are many different levels of help available to users, some of which have been explained above. In order from most explicit to most general:
?functionname at the
prompt.The Training committee recently published Learning and teaching biological data science in the Bioconductor community (Drnevich, Tan et al. 2025). It walks through how to get into Bioconductor depending on your goals and background, using or developing packages, efforts related to teaching Bioconductor, the computational infrastructure, language translation efforts, and minor recommendations on using AI/LLMs for teaching.