R Programming for Bioinformatics

Chapter 2

R Language Fundamentals

2.1 Introduction

In this chapter we introduce the basic language data types and discuss

their capabilities and structures. Then topics such as flow-control, iteration,

subsetting and exception handling will be presented. R directly supports two

different object-oriented programming (OOP) paradigms, which are discussed

in detail in Chapter 3. Many operations in R are vectorized, and understand-

ing and using vectorization is an essential component of becoming a proficient

programmer.

The R language was primarily designed as a language for data manipula-

tion, modeling and visualization, and many of the data structures reflect this

view. However, R is itself a full-fledged programming language, with its own

idioms – much like any other programming language. In some ways R can b e

considered as a functional programming language, although it is not purely

functional. R supports a form of lexical scope that provides a useful paradigm

for encapsulating computations.

R is an implementation of the S language (Becker et al., 1988; Chambers and

Hastie, 1992; Chambers, 1998). There is another commercial implementation

available from Insightful Corporation, called S-PLUS. The two implementa-

tions are quite similar, and much of the material covered here can be used in

either. However, there are many R-specific extensions that are used in this

monograph and users of R are our intended audience.

2.1.1 A brief introduction to R

We presume a reasonable familiarity with R but there are a few points that

will help to clarify some of the discussion. When R is started, a workspace is

created and that workspace is where the user creates and manipulates vari-

ables. This workspace is an environment, and an environment is a set of

bindings of names, or symbols, to values. The top-level workspace can be

accessed through its name, which is .GlobalEnv.

Assignment of value to a variable is generally done with either the = (equals)

character, or a special symb ol that is the concatenation of less than and mi-

nus, <-. Assignment creates a binding b etween a symbol and a value, in a

5

6 R Programming for Bioinformatics

particular environment. Removal of bindings is done with the function rm. In

the next code chunk, we create a symbol x and assign to it the value 10. We

then create a second symbol and assign the same value as x has.

> x = 10

> y = x

The value associated with y is a copy of the value asso ciated with x, and

changes to x do not affect y.

The semantics of rm(x) are that the association between x and its value

is broken and the symbol x is removed from the environment, but nothing is

done to the value that x referred to. If this value can b e acce ssed in other

ways, it will remain available. We provide an example in Section 2.2.4.3.

Valid variable names, sometimes referred to as syntactic names, are any

sequence of letters, digits, the period and the underscore, but they cannot

b egin with a digit or the underscore. If they begin with a period, the second

character cannot be a digit. Variable names that violate these rules must be

quoted (see the Quotes manual page) and the preferred quote is the backtick.

> _foo = 10

> "10:10" = 20

> ls()

[1] "10:10" "Rvers" "_foo" "basename"

[5] "biocUrls" "repos" "x" "y"

2.1.2 Attributes

Attributes can be attached to any R object except NULL and they are used

quite extensively. Attributes are stored, by name, in a list. All attributes can

b e retrieved using attributes, or any particular attribute can be accessed or

modified using the attr function. Attributes can be used by programmers to

attach any sort of information they want to any R object. R uses attributes

for many things; the S3 class system is based largely on attributes, dimensions

of arrays, and names on vectors, to name but a few.

In the code below, we attach an attribute to x and then show how the

printing of x changes to reflect the fact that it has an attribute.

> x = 1:10

> attr(x, "foo") = 11

> x