The keyboard evolver described at QwertAsdf determines that this is the optimal keyboard layout for the corpus of Lojban text I put into it (the same one I used for the word frequency lists):
W H Y S F G V ' N M / =
I O A E U D C L Z J -
Q , ; B . P T K R X
(To be clear, the evolver doesn't mess with the / = - keys, but they're a useful point of reference in looking at the layout.)
Even making slight changes to what is included in the corpus, the same layout comes out. Actually, there are variations which get the same score, by rearranging Q, W, and H. This makes a great deal of sense.
But there's something strange. T really is much more common than Z, yet Z ends up on the home row. This might have to do with the optimal locations for hitting combinations of consonants...