apply statement to sample columns, across rows of different lengths - r

Question

I'm trying to write a simple R function to sample 5-element substrings across two columns of a single data frame. The length of the strings are equal for each row, but they differ down the columns. The function works when I specify a row and col to act on, but I can't get the apply statement to work on on each row and each column. As written, it will only pull random samples based on the length of the first instance, so if the first instance is shorter than any of the other strings, the output for the other rows is sometimes less than 5-elements.
example df:
BP TF
1 CGTCTCTATTCTAGGCAAGA TTTFFFFTFFFTFFFTFTTT
2 AAGTCACTCGAATTCGGATGCCCCCTAGGC TTFFFFFTFFFFTTFTFFTTTFTTTTFTFF
3 TGCTCATGACGGGAC FFFTFTFFFFTFTFT
'intended output:'
1 CTATT FFTFF
2 CCTAG TTTFT
3 TCATG TFTFF
'reproducible example code:'
#make fake data frame
BaseP1 <- paste(sample(size = 20, x = c("A","C","T","G"), replace = TRUE), collapse = "")
BaseP2 <- paste(sample(size = 30, x = c("A","C","T","G"), replace = TRUE), collapse = "")
BaseP3 <- paste(sample(size = 15, x = c("A","C","T","G"), replace = TRUE), collapse = "")
TrueFalse1 <- paste(sample(size = 20, x = c("T","F"), replace = TRUE), collapse = "")
TrueFalse2 <- paste(sample(size = 30, x = c("T","F"), replace = TRUE), collapse = "")
TrueFalse3 <- paste(sample(size = 15, x = c("T","F"), replace = TRUE), collapse = "")
my_df <- data.frame(c(BaseP1,BaseP2,BaseP3), c(TrueFalse1, TrueFalse2, TrueFalse3))
Fragment = function(string) {
nStart = sample(1:nchar(string) -5, 1)
substr(string, nStart, nStart + 4)
}
Fragment(string = my_df[1,1])#works for the first row, first col.
but this does not work:
apply(my_df, c(1,2), function(x) Fragment(string = my_df[1:nrow(my_df),1:ncol(my_df)]))

Answer 1

There was an error in your function:
Fragment = function(string) {
nStart = sample(1:(nchar(string) -5), 1)
substr(string, nStart, nStart + 4)
}
It was missing parentheses between nchar(string) - 5, which made the subsetting go wrong.
You can then simply use apply(my_df, c(1,2), Fragment) as suggested in the comments.
To show that this works now:
for(i in 1:10000){
stopifnot(all(5 == sapply(apply(my_df, c(1,2), Fragment), nchar)))
}
This shows that in 10000 tries, it always produced 5 characters as output.