What I would like to see started is a seriously long-term experiment on mice (or something that is equally "visible" - fruit flies no good!).
Set up your control group of mice, and your experiment group. And in the experimental group, do some seriously active selection for particular physical traits (much as dog breeders have been doing for generations). Mice being mice, with fast reproduction, you should be able to push traits a lot further, a lot faster, than with dogs. Make the experimental group's lair conditions dramatically different to what a "normal" mouse would experience.
I'm talking a seriously long-term experiment here - 30+ years at a minimum - with the intention of proving that given enough generations, you will produce a creature significantly different to the mice in the control group.