Divide colums by other columns and itself depending on index in dplyr - r

Question

library(dplyr)
set.seed(1)
df <- data.frame(dddt_a = sample(1:1000, 1000, replace=T),
dddt_b = sample(1:1000, 1000, replace=T),
dddt_c = sample(1:1000, 1000, replace=T),
dddt_d = sample(1:1000, 1000, replace=T),
index = as.character(sample(c("a", "b"), 1000, replace=T)))
I want to divide each colum by either dddt_a or dddt_b depending on what the index is. If the index is a then divide all columns except the index by dddt_a and if index==b divide all columns except the index by dddt_b. The way it is set up now, this only divides dddt_a by a but not the other columns (likewise if index==b).
df1 <- df %>%
mutate_at(.vars = vars(starts_with("dddt")),
.funs = list(~ifelse(index=="a", ./dddt_a, ./dddt_b)))
head(df1)
dddt_a dddt_b dddt_c dddt_d index
1 1.0000000 686 474 756 a
2 0.7388466 1 681 726 b
3 1.0000000 218 570 448 a
4 2.0086393 1 830 958 b
5 1.0000000 989 590 128 a
6 1.0000000 128 978 144 a
A work around is storing the denominator variable outside, split the data for each index, divide everything and put it back together (I ran it only for index==a here). However, this should be possible in dplyr, I'm sure...?
ind_a <- df$dddt_a[df$index=="a"]
dfa <- df %>%
filter(index=="a")%>%
mutate_at(.vars = vars(starts_with("dddt")),
.funs = ~ ./!!ind_a)
Related to what seems to be the same issue. In a nex step I want to sum the values up, again depending on the index variable:
df2 <- df1 %>%
mutate(SUMS = ifelse(index=="a",
1+dddt_b+dddt_c+dddt_d,
1+dddt_a+dddt_c+dddt_d))
However, this sums all variables up...
head(df2)
dddt_a dddt_b dddt_c dddt_d index SUMS
1 1.0000000 686 474 756 a 1917.000
2 0.7388466 1 681 726 b 1408.739
3 1.0000000 218 570 448 a 1237.000
4 2.0086393 1 830 958 b 1791.009
5 1.0000000 989 590 128 a 1708.000
6 1.0000000 128 978 144 a 1251.000
But for the first row, for example, SUMS should be equal to 1916:
rowSums(df2[1,2:4]) #the result should be 1916 not 1917
1916
Thanks for the help.

Answer 1

Create a new column after dividing
library(dplyr)
df %>%
mutate_at(vars(starts_with("dddt")),
list(new = ~ifelse(index=="a", ./dddt_a, ./dddt_b))) %>%
head
# dddt_a dddt_b dddt_c dddt_d index dddt_a_new dddt_b_new dddt_c_new dddt_d_new
#1 836 686 474 756 a 1.000 0.821 0.567 0.904
#2 679 919 681 726 b 0.739 1.000 0.741 0.790
#3 129 218 570 448 a 1.000 1.690 4.419 3.473
#4 930 463 830 958 b 2.009 1.000 1.793 2.069
#5 509 989 590 128 a 1.000 1.943 1.159 0.251
#6 471 128 978 144 a 1.000 0.272 2.076 0.306
If you want you can then select only "_new" columns or rename the "_new" column to names of your choice.

Answer 2

We can also use case_when
library(dplyr)
df %>%
mutate_at(vars(starts_with("dddt")),
list(new = ~case_when(index=="a" ~ ./dddt_a, TRUE ~ ./dddt_b)))