Say I have a vector named all_combinations with numbers from 1 to 20.
I need to extract 2 vectors (coding_1 and coding_2) of length equal to number_of_peptide_clusters, which happens to be 20 as well in my current case.
The 2 new vectors should be randomly sampled from all_combinations, so that are not overlapping at each index position.
I do the following:
set.seed(3)
all_combinations=1:20
number_of_peptide_clusters=20
coding_1 <- sample(all_combinations, number_of_peptide_clusters, replace = FALSE)
coding_1
[1] 5 12 7 4 10 8 11 15 17 16 18 13 9 20 2 14 19 1 3 6
coding_2 <- sample(all_combinations, number_of_peptide_clusters, replace = FALSE)
coding_2
[1] 5 9 19 16 18 12 8 6 15 3 13 14 7 2 11 20 10 4 17 1
This is the example that gives me trouble, cause only one number is overlapping at the same index (5 at position 1).
What I would do in these cases is spot the overlapping numbers and resample them out of the list of all overlapping numbers...
Imagine coding_1 and coding_2 were:
coding_1
[1] 5 9 7 4 10 8 11 15 17 16 18 13 12 20 2 14 19 1 3 6
coding_2
[1] 5 9 19 16 18 12 8 6 15 3 13 14 7 2 11 20 10 4 17 1
In this case I would have 5 and 9 overlapping in the same position, so I would resample them in coding_2 out of the full list of overlapping ones [resample index 1 from c(5,9) so that isn't equal to 5, and index 2 so it isn't equal to 9]. So coding_2 would be:
coding_2
[1] 9 5 19 16 18 12 8 6 15 3 13 14 7 2 11 20 10 4 17 1
However, in the particular case above, I cannot use such approach... So what would be the best way to obtain 2 samples of length 20 from a vector of length 20 as well, so that the samples aren't overlapping at the same index positions?
It would be great that I could obtain the second sample coding_2 already knowing coding_1... Otherwise obtaining the 2 at the same time would also be acceptable if it makes things easier. Thanks!
I think the best solution is simply to use a rejection strategy:
set.seed(3)
all_combinations <- 1:20
number_of_peptide_clusters <- 20
count <- 0
repeat {
count <- count + 1
message("Try number ", count)
coding_1 <- sample(all_combinations, number_of_peptide_clusters, replace = FALSE)
coding_2 <- sample(all_combinations, number_of_peptide_clusters, replace = FALSE)
if (!any(coding_1 == coding_2))
break
}
#> Try number 1
#> Try number 2
#> Try number 3
#> Try number 4
#> Try number 5
#> Try number 6
#> Try number 7
#> Try number 8
#> Try number 9
coding_1
#> [1] 18 16 17 12 13 8 6 15 3 5 20 9 11 4 19 2 14 7 1 10
coding_2
#> [1] 5 20 14 2 11 6 7 10 19 8 4 1 15 9 13 17 18 16 12 3
Created on 2020-11-04 by the reprex package (v0.3.0)