Intermediate programming with R

Conditional statements

Learning Objectives

  • Filter using logical vectors created with conditional statements
  • Search for patterns with grepl
  • Make decisions with if and else statements

In the previous lesson, we were introduced to logical vectors with the functions is.na and anyNA.

counts_raw$authorsCount[1:10]
 [1]  6 14 NA NA  6 10 NA NA NA  5

is.na(counts_raw$authorsCount[1:10])
 [1] FALSE FALSE  TRUE  TRUE FALSE FALSE  TRUE  TRUE  TRUE FALSE

anyNA(counts_raw$authorsCount[1:10])
[1] TRUE

In this lesson we will learn how these types of logical vectors can be used for filtering data and making decisions.

Filtering with logical vectors

Instead of providing the numbers of the rows we want, we can filter with a logical vector.

counts_raw$authorsCount[1:10]
 [1]  6 14 NA NA  6 10 NA NA NA  5

counts_raw$authorsCount[1:10] > 7
 [1] FALSE  TRUE    NA    NA FALSE  TRUE    NA    NA    NA FALSE

dim(counts_raw[counts_raw$authorsCount > 7, ])
[1] 10348    32

Here we filtered the data to only include the 10348 rows where the number of authors was greater than 7.

To filter for equality or non-equality, use == or !=:

# All the articles published in the journal PLOS One
dim(counts_raw[counts_raw$journal == "pone", ])
[1] 14099    32

# All the articles NOT published in the journal PLOS One
dim(counts_raw[counts_raw$journal != "pone", ])
[1] 10232    32

Here are the other possibilities:

  • > - “greater than”
  • < - “less than”
  • >= - “greater than or equal to”
  • <= - “less than or equal to”
  • == - “equal to”
  • != - “not equal to”

These logical conditions can be combined into more complex filters using the ampersand & (“and”) or vertical bar | (“or”) operators.

# All the articles published in the journal PLOS One AND with more than 7 authors
dim(counts_raw[counts_raw$journal == "pone" &
               counts_raw$authorsCount > 7, ])
[1] 4697   32

# All the articles published in the journal PLOS One OR the journal PLOS Biology
dim(counts_raw[counts_raw$journal == "pone" |
               counts_raw$journal == "pbio", ])
[1] 16690    32

When we are checking one vector for multiple possibilities, it is more convenient to use the operator %in% instead of creating multiple “or” conditions.

# All the articles published in the journals PLOS One, PLOS Biology, or PLOS Genetics
dim(counts_raw[counts_raw$journal %in% c("pone", "pbio", "pgen"), ])
[1] 18459    32

Lastly, to reverse any logical vector, we can append the exclamation point ! for “NOT”.

# All the articles NOT published in the journals PLOS One, PLOS Biology, or PLOS Genetics
dim(counts_raw[!(counts_raw$journal %in% c("pone", "pbio", "pgen")), ])
[1] 5872   32

Finding patterns with grepl

We saw in the Unix shell that we could search for lines in a file that contain a specific pattern using grep. R provides similar functionality. grepl searches each element of a vector for a given pattern and returns TRUE if it finds it, and FALSE otherwise. Let’s try it out using the column plosSubjectTags, which describes the scientific discipline(s) of the article.

head(counts_raw$plosSubjectTags)
[1] Cell Biology|Immunology|Molecular Biology                                             
[2] Biotechnology|Genetics and Genomics|Infectious Diseases|Virology                      
[3] Computational Biology|Biotechnology|Genetics and Genomics|Infectious Diseases|Virology
[4] Cell Biology|Immunology|Molecular Biology                                             
[5] Genetics and Genomics|Infectious Diseases|Microbiology                                
[6] Ecology|Evolutionary Biology|Genetics and Genomics                                    
6715 Levels: Anesthesiology and Pain Management ...

How many of the articles have to do with “Immunology”?

dim(counts_raw[grepl("Immunology", counts_raw$plosSubjectTags), ])
[1] 2708   32

The first argument grepl was the string we were searching for, and the second argument was the vector to be searched.

How many of the immunology articles were published in PLOS Medicine.

dim(counts_raw[grepl("Immunology", counts_raw$plosSubjectTags) &
                 counts_raw$journal == "pmed", ])
[1] 194  32

Making decisions

In addition to filtering, we can use conditional statements to adapt the behavior of the code based on the input data. We do this using if and else statements. The basic structure is the following:

if (condition is TRUE) {
  do something
} else {
  do a different thing
}

For example, we can check whether a vector contains any missing values.

x <- counts_raw$authorsCount
if (anyNA(x)) {
  print("Be careful! The data contains missing values.")
} else {
  print("Looks good. The data does NOT contain missing values.")
}
[1] "Be careful! The data contains missing values."

Or we can check if an object is a specific data type, and convert it to the one we need. Here we convert the column title from a factor to a character vector.

x <- counts_raw$title
if (!is.character(x)) {
  x <- as.character(x)
}

Challenges

Filtering articles

How many articles with the subject tag (plosSubjectTags) “Evolutionary Biology” were published in either PLOS One (“pone”), PLOS Biology (“pbio”), or PLOS Medicine (“pmed”)?

Software Carpentry Source Contact License