Improve the performance of this script - r

Question

Here is a piece of my code that I introduce in an R shiny application but which takes me a lot of time because I execute it in a reactive function which I then call five times for different graphics.
Do you have an idea to improve the speed of this script?
I have already tried to execute this with purr but I do not master this tool well enough.
Here is a reproducible example
library(profvis)
profvis({
#dataframe created for the example
DF<- data.frame("scan"=seq(1:7518),"dye1"=NA,"dye2"=NA,"dye3"=NA,"dye4"=NA,"dye5"=NA,"dye6"=NA)
DF$dye1 <- sample(100, size = nrow(DF), replace = TRUE)
DF$dye2 <- sample(100, size = nrow(DF), replace = TRUE)
DF$dye3 <- sample(100, size = nrow(DF), replace = TRUE)
DF$dye4 <- sample(100, size = nrow(DF), replace = TRUE)
DF$dye5 <- sample(100, size = nrow(DF), replace = TRUE)
DF$dye6 <- sample(100, size = nrow(DF), replace = TRUE)
#slowness begins here
for (d in 3000:7518){
#array of input data
input <- numeric(1206)
for (i in -100:100){
input[1+i+100] <- DF$dye1[d + i]
input[202+i+100] <- DF$dye2[d + i]
input[403+i+100] <- DF$dye3[d + i]
input[604+i+100] <- DF$dye4[d + i]
input[805+i+100] <- DF$dye5[d + i]
input[1006+i+100] <- DF$dye6[d + i]
}
}
})

Answer 1

First: please really reconsider what you want to achieve and whether this approach is the smartest way to achieve it..
Second: use vectorization to improve your performance:
d <- 3000
input <- numeric(1206)
microbenchmark::microbenchmark(
# loop as before
case1 = {for (i in -100:100){
input[1+i+100] <- DF$dye1[d + i]
}},
# use vectorization
case2 = {input[(1-100+100):(1+100+100)] <- DF$dye1[(d -100):(d +100)]}
)