gsub inside function with mutate called in dplyr chain gives error - r

Question

I have the following example data (the real data contains other columns with both numeric and character variables):
structure(list(AM = structure(1:20, .Label = c("AMP_R", "AZI_R",
"CHL_R", "CIP_R", "COL_R", "ERY_R", "ETP_R", "F.C_R", "FEP_R",
"FOT_R", "FOX_R", "GEN_R", "IMI_R", "MERO_R", "NAL_R", "STR_R",
"SULFA_R", "T.C_R", "TAZ_R", "TET_R"), class = "factor")), .Names = "AM", row.names = c(NA,
-20L), class = "data.frame")
I tried to create a function that will detect whether or not a column in a data frame contains variables with the ending "_R". If they do, it will remove this ending and proceed with renaming the variables to full names, accoring to a conversion table. If the "_R" ending is not present, it will just convert the names directly.
I have tried the following on the first part of the function:
library(dplyr)
convert_AM_names <- function(data, col) {
data %>%
mutate(col = gsub("(.*?)_R", "\\1", col))
}
I want to use it in a dplyr chain, like this:
AM <- AM %>%
rowwise() %>%
convert_AM_names(., AM)
However, when I do this, it gives the error "Error in mutate_impl(.data, dots): Column "col" must be length 1, not 20"
I saw that similar issues have been addressed here at SO, but for most of them the solution was to use rowwise(), which doesn't seem to work here. Any suggestions?

Answer 1

You can use an anchor for your regular expression that only matches when the _R is right at the end:
convert_AM_names <- function(col) {
gsub("(.*)_R$", "\\1", col)
}
library(dplyr)
df %>%
mutate(AM = convert_AM_names(AM))
Or directly - without the overhead of convert_AM_names():
df %>%
mutate(AM = gsub("(.*)_R$", "\\1", AM))
Both will yield:
AM
1 AMP
2 AZI
3 CHL
4 CIP
5 COL
6 ERY
7 ETP
8 F.C
9 FEP
10 FOT
11 FOX
12 GEN
13 IMI
14 MERO
15 NAL
16 STR
17 SULFA
18 T.C
19 TAZ
20 TET

Answer 2

You can use mutate_at() which allows you to select a column and apply a function to it.
AM %>%
mutate_at(.vars = "AM",
.funs = gsub,
pattern = "(.*?)_R",
replacement = "\\1")
If you wanted, you could also rewrite your function:
convert_AM_names <- function(col) {
gsub("(.*?)_R", "\\1", col)
}
And use it in mutate_at():
AM %>%
mutate_at(.vars = "AM",
.funs = convert_AM_names)
In both cases, the result looks like this:
AM
1 AMP
2 AZI
3 CHL
4 CIP
5 COL
6 ERY
7 ETP
8 F.C
9 FEP
10 FOT
11 FOX
12 GEN
13 IMI
14 MERO
15 NAL
16 STR
17 SULFA
18 T.C
19 TAZ
20 TET