R Programming for Bioinformatics

Chapter 7

R Packages

An R package typically consists of a coherent collection of functions and data

structures that are suitable for addressing a particular problem. Packages

are easy to distribute and share with others, and in many ways writing a

package is the best and most effective way to share your software and ideas

with others. Learning to move away from collections of scripts, or functions,

that get sourced to software organized in packages is very enabling. Writing

your first package can be a challenge, but the second and third are simpler.

There are many advantages to writing a package, including the fact that

everyone you collaborate with can be using precisely the same code and all

can contribute to its quality and usability.

In many cases, developers will either have access to existing co de written

in some other language, or they may find that they need to implement some

parts of their software in C, or some other compilable language, for efficiency.

This topic is covered in Chapter 6.

The many hundreds of contributed packages provide a great resource, and

for many problems you will simply want to find, download, install and use

packages written by others. There are currently three main repositories for

R packages: CRAN, Bioconductor and Omegahat. CRAN contains the most

R packages (over 1000) while Bioconductor and Omegahat are smaller. The

number of packages is somewhat overwhelming and two related innovations,

CRAN Task Views (the ctv package) and the biocViews packages, provide

to ols that can be used to provide overviews of packages suitable for different

subject areas. Both approaches use restricted sets of terms to help orga-

nize packages. We will discuss in Section 7.2.1 some of the ways in which

biocViews can be used and readers who visit the Bioconductor web site will

see it in action, as the biocViews terms are used to organize the Bioconductor

packages.

In this chapter we first review some of the available functionality for using

packages and then discuss how to write your own packages. For most of the

topics we will discuss, the R Extensions Manual (R Development Core Team,

2007c) is a more comprehensive reference.

211