AFAICT, they only differ in being zero- vs. one-based, i.e. in my output "4" means 4th character. Both can be trivially converted to one another by adding or subtracting 1.  Or which difference are you referring to?

in some very rare cases i have a conversion from an A->N. the N is just another character. the other code snippets catch this type of conversion whereas your code calls it a G->A. why is this the case?

That's because you didn't mention the "N" in your original post :)

The idea of the transliteration is that the XOR value computed for every (directed) comparison of characters is different. This can only be determined for a predefined set of allowed characters.

To also allow "N", you could (for example) use the transliteration tr/ATCGN/J4XD7/. With this, the XOR values for the respective changes would compute as:

       XOR val    change

          \x0b => A->A  *    ( "A" ^ "J" )
          \x19 => A->C       ( "A" ^ "X" )
          \x05 => A->G       ( "A" ^ "D" )
          \x76 => A->N       ...
          \x75 => A->T
          \x09 => C->A
          \x1b => C->C  *
          \x07 => C->G
          \x74 => C->N
          \x77 => C->T
          \x0d => G->A
          \x1f => G->C
          \x03 => G->G  *
          \x70 => G->N
          \x73 => G->T
          \x04 => N->A
          \x16 => N->C
          \x0a => N->G
          \x79 => N->N  *
          \x7a => N->T
          \x1e => T->A
          \x0c => T->C
          \x10 => T->G
          \x63 => T->N
          \x60 => T->T  *
[download]

The ones marked with "*" are the "no-changes", which should make up the exclusion character set in the final match. I.e., with the above modified transliteration, you should change that to

    while ($diff =~ /([^\x0b\x1b\x03\x79\x60])/g) {
[download]