what you/we really want is an implementation of the below - an optimal way to approximately match regular expressions. why this rather than the others above?

1: Bull Math Biol. 1989;51(1):5-37.

Approximate matching of regular expressions.

Myers EW, Miller W.
[download]

none of the above are able to compare "edit distances" *for regular expressions* in the way the Text::Levenshtein etc allow the comparison of these edit distances for strings. instead, they quite effectively hardwire a greater degree of flexibility into the patterns that can be recognized. but to do this properly, you need to 'penalize' insertions/deletions in your regexp in the same way you do for sequences. the above paper outlines a way of doing this. as for implementation - I don't know.

is there something around the BioPerl guys might know of?