Numbers Within Strings

2022-07-24

A common way to encode numerical data

It’s common for a lot of numerical information to be encoded in strings, particularly in file names. Consider a series of microscope images of cells from different patients detailing the patient number, the cell number and the number of hours after biopsy that the image was taken. They might be named like:

img_names
#>  [1] "patient1-cell1-0hours-after-biopsy.tif"  
#>  [2] "patient1-cell1-2.5hours-after-biopsy.tif"
#>  [3] "patient1-cell2-0hours-after-biopsy.tif"  
#>  [4] "patient1-cell2-2.5hours-after-biopsy.tif"
#>  [5] "patient1-cell3-0hours-after-biopsy.tif"  
#>  [6] "patient1-cell3-2.5hours-after-biopsy.tif"
#>  [7] "patient2-cell1-0hours-after-biopsy.tif"  
#>  [8] "patient2-cell1-2.5hours-after-biopsy.tif"
#>  [9] "patient2-cell2-0hours-after-biopsy.tif"  
#> [10] "patient2-cell2-2.5hours-after-biopsy.tif"
#> [11] "patient2-cell3-0hours-after-biopsy.tif"  
#> [12] "patient2-cell3-2.5hours-after-biopsy.tif"

All of the numbers

For some crude reason, you might just want all of the numbers:

library(strex)
str_extract_numbers(img_names)
#> [[1]]
#> [1] 1 1 0
#> 
#> [[2]]
#> [1] 1 1 2 5
#> 
#> [[3]]
#> [1] 1 2 0
#> 
#> [[4]]
#> [1] 1 2 2 5
#> 
#> [[5]]
#> [1] 1 3 0
#> 
#> [[6]]
#> [1] 1 3 2 5
#> 
#> [[7]]
#> [1] 2 1 0
#> 
#> [[8]]
#> [1] 2 1 2 5
#> 
#> [[9]]
#> [1] 2 2 0
#> 
#> [[10]]
#> [1] 2 2 2 5
#> 
#> [[11]]
#> [1] 2 3 0
#> 
#> [[12]]
#> [1] 2 3 2 5

It seems to have missed the fact that 2.5 is a number and not two numbers 2 and 5. This is because the default is decimals = FALSE. To recognise decimals, set decimals = TRUE. Also, note that there is an option to recognise scientific notation. More on that below.

str_extract_numbers(img_names, decimals = TRUE)
#> [[1]]
#> [1] 1 1 0
#> 
#> [[2]]
#> [1] 1.0 1.0 2.5
#> 
#> [[3]]
#> [1] 1 2 0
#> 
#> [[4]]
#> [1] 1.0 2.0 2.5
#> 
#> [[5]]
#> [1] 1 3 0
#> 
#> [[6]]
#> [1] 1.0 3.0 2.5
#> 
#> [[7]]
#> [1] 2 1 0
#> 
#> [[8]]
#> [1] 2.0 1.0 2.5
#> 
#> [[9]]
#> [1] 2 2 0
#> 
#> [[10]]
#> [1] 2.0 2.0 2.5
#> 
#> [[11]]
#> [1] 2 3 0
#> 
#> [[12]]
#> [1] 2.0 3.0 2.5

It’s also possible to extract the non-numeric parts of the strings:

str_extract_non_numerics(img_names, decimals = TRUE)
#> [[1]]
#> [1] "patient"                "-cell"                  "-"                     
#> [4] "hours-after-biopsy.tif"
#> 
#> [[2]]
#> [1] "patient"                "-cell"                  "-"                     
#> [4] "hours-after-biopsy.tif"
#> 
#> [[3]]
#> [1] "patient"                "-cell"                  "-"                     
#> [4] "hours-after-biopsy.tif"
#> 
#> [[4]]
#> [1] "patient"                "-cell"                  "-"                     
#> [4] "hours-after-biopsy.tif"
#> 
#> [[5]]
#> [1] "patient"                "-cell"                  "-"                     
#> [4] "hours-after-biopsy.tif"
#> 
#> [[6]]
#> [1] "patient"                "-cell"                  "-"                     
#> [4] "hours-after-biopsy.tif"
#> 
#> [[7]]
#> [1] "patient"                "-cell"                  "-"                     
#> [4] "hours-after-biopsy.tif"
#> 
#> [[8]]
#> [1] "patient"                "-cell"                  "-"                     
#> [4] "hours-after-biopsy.tif"
#> 
#> [[9]]
#> [1] "patient"                "-cell"                  "-"                     
#> [4] "hours-after-biopsy.tif"
#> 
#> [[10]]
#> [1] "patient"                "-cell"                  "-"                     
#> [4] "hours-after-biopsy.tif"
#> 
#> [[11]]
#> [1] "patient"                "-cell"                  "-"                     
#> [4] "hours-after-biopsy.tif"
#> 
#> [[12]]
#> [1] "patient"                "-cell"                  "-"                     
#> [4] "hours-after-biopsy.tif"

Extract specific numbers

What if we just want the cell number from each image?

The nth number

We know the cell number is always the second number, so we can use the str_nth_number() function with n = 2.

str_nth_number(img_names, n = 2)
#>  [1] 1 1 2 2 3 3 1 1 2 2 3 3

Numbers after patterns

To be more specific, you could say the cell number is the first number after the first instance of the word “cell”. To go this route, strex provides str_nth_number_after_mth() which gives the nth number after the mth appearance of a given pattern:

str_nth_number_after_mth(img_names, "cell", n = 1, m = 1)
#>  [1] 1 1 2 2 3 3 1 1 2 2 3 3

There’s also a convenient wrapper for getting the first number after the first appearance of a pattern:

str_first_number_after_first(img_names, "cell")
#>  [1] 1 1 2 2 3 3 1 1 2 2 3 3

Numbers before patterns

Now what if we want the number of hours after biopsy for each image? Looking at the image file names, we’d need the last number before the first occurrence of the word “biopsy”.

str_last_number_before_first(img_names, "biopsy", decimals = TRUE)
#>  [1] 0.0 2.5 0.0 2.5 0.0 2.5 0.0 2.5 0.0 2.5 0.0 2.5

Tidy number extraction

To extract all of this information tidily, use a data frame:

data.frame(img_names,
  patient = str_first_number_after_first(img_names, "patient"),
  cell = str_first_number_after_first(img_names, "cell"),
  hrs_after_biop = str_last_number_before_first(img_names, "biop",
    decimals = TRUE
  )
)
#>                                   img_names patient cell hrs_after_biop
#> 1    patient1-cell1-0hours-after-biopsy.tif       1    1            0.0
#> 2  patient1-cell1-2.5hours-after-biopsy.tif       1    1            2.5
#> 3    patient1-cell2-0hours-after-biopsy.tif       1    2            0.0
#> 4  patient1-cell2-2.5hours-after-biopsy.tif       1    2            2.5
#> 5    patient1-cell3-0hours-after-biopsy.tif       1    3            0.0
#> 6  patient1-cell3-2.5hours-after-biopsy.tif       1    3            2.5
#> 7    patient2-cell1-0hours-after-biopsy.tif       2    1            0.0
#> 8  patient2-cell1-2.5hours-after-biopsy.tif       2    1            2.5
#> 9    patient2-cell2-0hours-after-biopsy.tif       2    2            0.0
#> 10 patient2-cell2-2.5hours-after-biopsy.tif       2    2            2.5
#> 11   patient2-cell3-0hours-after-biopsy.tif       2    3            0.0
#> 12 patient2-cell3-2.5hours-after-biopsy.tif       2    3            2.5

Other number formats

strex can also deal with numbers in scientific and comma notation.

string <- c("$1,000", "$1e6")
str_first_number(string, commas = TRUE, sci = TRUE)
#> [1] 1e+03 1e+06

All of the number functions

There are a whole host of functions for extracting numbers from strings in the strex package:

str_subset(ls("package:strex"), "number")
#>  [1] "str_extract_numbers"           "str_first_number"             
#>  [3] "str_first_number_after_first"  "str_first_number_after_last"  
#>  [5] "str_first_number_after_mth"    "str_first_number_before_first"
#>  [7] "str_first_number_before_last"  "str_first_number_before_mth"  
#>  [9] "str_last_number"               "str_last_number_after_first"  
#> [11] "str_last_number_after_last"    "str_last_number_after_mth"    
#> [13] "str_last_number_before_first"  "str_last_number_before_last"  
#> [15] "str_last_number_before_mth"    "str_nth_number"               
#> [17] "str_nth_number_after_first"    "str_nth_number_after_last"    
#> [19] "str_nth_number_after_mth"      "str_nth_number_before_first"  
#> [21] "str_nth_number_before_last"    "str_nth_number_before_mth"    
#> [23] "str_split_by_numbers"

Regular expression

Of course, all of the above is possible with regular expression using stringr, it’s just more difficult and less expressive:

data.frame(img_names,
  patient = str_match(img_names, "patient(\\d+)")[, 2],
  cell = str_match(img_names, "cell(\\d+)")[, 2],
  hrs_after_biop = str_match(img_names, "(\\d*\\.*\\d+)hour")[, 2]
)
#>                                   img_names patient cell hrs_after_biop
#> 1    patient1-cell1-0hours-after-biopsy.tif       1    1              0
#> 2  patient1-cell1-2.5hours-after-biopsy.tif       1    1            2.5
#> 3    patient1-cell2-0hours-after-biopsy.tif       1    2              0
#> 4  patient1-cell2-2.5hours-after-biopsy.tif       1    2            2.5
#> 5    patient1-cell3-0hours-after-biopsy.tif       1    3              0
#> 6  patient1-cell3-2.5hours-after-biopsy.tif       1    3            2.5
#> 7    patient2-cell1-0hours-after-biopsy.tif       2    1              0
#> 8  patient2-cell1-2.5hours-after-biopsy.tif       2    1            2.5
#> 9    patient2-cell2-0hours-after-biopsy.tif       2    2              0
#> 10 patient2-cell2-2.5hours-after-biopsy.tif       2    2            2.5
#> 11   patient2-cell3-0hours-after-biopsy.tif       2    3              0
#> 12 patient2-cell3-2.5hours-after-biopsy.tif       2    3            2.5