[EMBOSS] how string is shuffleseq

[Derek Gatherer]

Dear EMBOSS experts

How good a shuffler is shuffleseq?  I have always believed that it 
shuffles beyond all recognition, since tracer words inserted into a 
database are always gone after 1 round of shuffling.  But are there 
any hard test data on its shuffling strength?  For instance, for any, 
say, 100-mer, what fragments would reasonably be expected to remain 
after shuffling?  Should I be surprised to see a 15-mer from the 
100-mer still intact??

Sorry I can't express this more precisely.  I thinks it's probably 
clear...  I ask as there is a piece of software I am testing on real 
genomes and their shuffled derivatives, and I want to exclude the 
possibility that the control is compromised by an inadequacy of shuffling.

Cheers
Derek