Looks fine to me, but the best way would be to verify it yourself, print out the before & after sequences below each other and change the window size to something readable (e.g. 5).
That said, I think the problem maybe miss-represented, are you sure you dont want to break up the DNA into window-sized chunks and shuffle the chunks? This would be a more useful exercise!
Update:
...SO VERY different from what has been published
It would be nice to see a link to the published work!
This is not a Signature...