I'm having trouble figuring out how to use a string containing R data frame column names to do some basic calculations to mutate into new columns. For example, I have columns of baseline values and other columns with post-treatment timepoints. I want to use strings of column names for this because I will be using data with different timepoints specified and I want a programmatic solution.
For example, I have this data frame, and I think I need to use some of the syntax in my mutate line below, but can't figure out exactly how to write the right hand side. I want columns called 'day1_fc' and 'day2_fc' to represent the fold change of day1/baseline, and day2/baseline.
df <- data.frame(day0 = c(1,1,1),
day1 = c(2,3,4),
day2 = c(3,4,5))
baseline = 'day0'
sym_baseline <- sym(baseline)
post = c('day1', 'day2')
post1 <- post[1]
post2 <- post[2]
df %>%
mutate(!!paste0(post1, '_fc' := ?????),
!!paste0(post2, '_fc') := ?????)
I want the result to look like:
df <- data.frame(day0 = c(1, 0.5, 2),
day1 = c(2, 3, 4),
day2 = c(3, 4, 5),
day1_fc = c(2, 6, 2),
day2_fc = c(3, 8, 2.5))
You can use :
library(dplyr)
library(rlang)
df %>%
mutate(!!paste0(post1, '_fc') := !!sym(post[1])/!!sym_baseline,
!!paste0(post2, '_fc') := !!sym(post[2])/!!sym_baseline)
# day0 day1 day2 day1_fc day2_fc
#1 1.0 2 3 2 3.0
#2 0.5 3 4 6 8.0
#3 2.0 4 5 2 2.5
A general solution for many values of post would be using map :
bind_cols(df, purrr::map_dfc(post,
~df %>% transmute(!!paste0(.x, '_fc') := !!sym(.x)/!!sym_baseline)))