4  Working with text patterns (regular expressions)

4.1 Introduction

Functions from the Tidyverse packages (dplyr and tidyr) will be used in this tutorial. As example data the lipid names of a lipidomics dataset is being.

library(here)
library(tidyverse)

mydata <- read_csv(file = here("data/Testdata_wide.csv")) |> names()

To selecting column contain specific text patterns use matches(). To select rows, replace and remove specific text in rows and cells use functions of the stringr package, which all start with str_. using fu This function uses regular expressions to match column names. A few key elements in regular expression strings:

  • To search for different strings: | (OR operator)
  • Starts with: ^ Ends with: $
  • Escape character: \\. Symbols as :|\/[] etc have a function in regular expression, if you want search for them you need add \\ in front
  • Selections [ ] are used to indicate any of the characters is allowed. Use - to indicate ranges, and
    ^ to exclude any of the characters. Examples: [abcd], [a-d], [1234], [1-4], [^SI]
  • Dot . indicates any character, \\d any digit, \\D any non-digit, \\s any whitespace, \\S any non-whitespace
  • Plus + means any, and {n} a defined number n of the preceding characters or symbols.
    Examples: [1-9]+, [1-9]{6}, \\d{6}, \\s+ 
# select PE and PC species (= match either PC or PE)
mydata |> str_subset("^PC|^PE")

# select species ending with :4 (e.g. CE 20:4). Need 'escape' the : using \\
mydata |> str_subset("\\:4$") 

# select PC, PE, PI, PS and PG. Match a P at start and a selection of C,E,I,S,G
mydata |> str_subset("^P[CEISG]")

# select all PC but not PC O- and PC P- (match PC start, a space, and a digit)
mydata |> str_subset("^PC \\d")

# select all Cer with with FA chains C14-C18
mydata |> str_subset("^Cer .+/1[4-8]")
mydata |> str_subset("^Cer d\\d{2}\\:[0-2]\\/1[4-8]")

To test or explore your regular expressions the stringr function str_view() can be helpful..

# Shows all elements with matched items highlighted
mydata |> str_view("^Cer d\\d{2}\\:[0-2]\\/1[4-8]", match = FALSE)

# Shows only matched items with matches characters highlighted
mydata |> str_view("^Cer d\\d{2}\\:[0-2]\\/1[4-8]", match = TRUE)