Mulling the genome and rapid speciation

Thought:

Variation through change in gene frequency as populations are isolated from other populations is such a common event in nature, occurring within observable time periods and not requiring millions of years, even on out to actual "speciation" as defined by evolutionists, not to mention that it's the way new domestic breeds are intentionally developed, it seems to me the question could be raised:

In all this shuffling and shifting of alleles and development of new phenotypes therefrom has any gene for anything other than variations on the basic structure of the Kind been observed to change? If macroevolution ever occurred wouldn't you expect that in all that shifting of alleles you'd see some alleles shift that pertain to the very dogness of dogs, the very catness of cats, the very horseness of horses? But we have huge dogs and tiny dogs and an enormous range of types in between and there is no doubt they are all dogs; same with cats.

Can they even distinguish genes that pertain to these things from others in the genome? Superficial traits such as eye color, skin color, hair and fur color and type, overall body size and musculature, these things are easily and rapidly changed over comparatively few generations. But has anyone ever seen a dog eye change in the direction of a cat eye or vice versa, or the yapping or barking of dogs change into something else etc. Or work with what evolutionism believes: could you breed a reptile of your choice into something that is even somewhat birdlike? (Since there were once apparently flying reptiles (in the pre-Flood world) I would expect you MIGHT be able to tap into that genetic possibility in a reptilian genome (maybe it's all gone to junk DNA now though) but could you get a scale to evolve into a feather?)

That is, has anything that is DEFINITIVE of a Kind ever shown the slightest change into something else from all this alteration in the genes and alleles over observable generations? Could it? It doesn't happen, does it?

I'm thinking the genome must obey some laws that haven't yet been discovered, that keep the basic blueprint of the Kind from changing while a large variety of secondary characteristics may change freely.