Packages & Libraries

A package is a collection of R functions, data, and code bundled together. A library is the directory where packages are stored on your computer and the default is R_HOME/library.

Types of Packages

1. Base packages: Comes with R installation & no installation is needed. For example, stats, utils, datasets, methods.
2. Recommended packages: Part of R distribution but need to be loaded explicitly. For example, MASS, boot, class, cluster.
3. Contributed packages: Created by R community available on CRAN, Bioconductor, GitHub etc. For example, ggplot2, dplyr.

Installing a package

install.packages(“xlsx”)

Getting info about installed packages

installed.packages() 

Installing multiple packages at once

install.packages(c(“tidyverse”, “xlsx”, “dlookr”))

Installing from Bioconductor

if (!require(“BiocManager”, quietly = TRUE))
      install.packages(“BiocManager”)

BiocManager::install(“GenomicFeatures”)

Installing from GitHub

install.packages(“devtools”)
devtools::install_github(“username/repository”)

Installing based on specific version

devtools::install_version(“dplyr”, version = “1.0.0”)

Loading package

library(xlsx)

Loading multiple packages

packages1 <- c(“tidyverse”, ”xlsx”, “dlookr”)
lapply(packages1, library, character.only = TRUE) or
lapply(c(“tidyverse”, ”xlsx”, “dlookr”), require, character.only = TRUE)

Checking if a particular package is installed

system.file(package=’ggplot2′)

Checking for a package version

packageVersion(“dplyr”)

Listing all loaded packages

sessionInfo()

Namespace management is an option when functions are same in different libraries.  Use the library name to specify the function clearly.

dplyr::select(mtcars, mpg)

Dependency conflicts in R occur when multiple packages require different versions of the same dependency. This can lead to errors and dysfunctional code.

Checking for any conflicts

conflicts()

Detaching a package

detach(“package:dplyr”)

Unloading all packages

detach(“package:dplyr”, unload = TRUE)

Saving all packages list for reproducibility

writeLines(capture.output(sessionInfo()), “session_info.txt”)

Updating all packages

update.packages()

Removing  packages

remove.packages(“dplyr”)
remove.packages(c(“dplyr”, “xlsx”))

Cleaning up unused dependencies

install.packages(“pacman”)
pacman::p_clean()

Repository management is the process of configuring and using servers to install packages. Set the

CRAN mirror to choose a location in install packages for speed (choose mirror closer to you), network (CRAN is a network of mirrors around the world, not just a website), cloud (use  cloud.r-project.org

service to automatically select a server). RStudio’s package manager works like a CRAN mirror, and

includes packages from Bioconductor as well as CRAN.

Listing all available CRAN mirrors

getCRANmirrors()

To set a mirror

options(repos = c(CRAN = “https://cloud.r-project.org”))

to get local libraries

.libPaths()

To add new library location

.libPaths(c(.libPaths(), “path/to/new/library”))

Best practices

1. Load packages at the start of scripts
2. Use explicit namespaces for clarity
3. Document dependencies