Ok to be honest though my gut feeling is that statistically making these sorts of inferences over 30 mutations seems pretty weak, especially since the level of

Ok to be honest though my gut feeling is that statistically making these sorts of inferences over 30 mutations seems pretty weak, especially since the level of difference between say human and mouse is not that great to start out with, and because these mutations are not independent; they are dependent on survival of the virus. I guess part of the problem is that I don't know how to do the statistics on this sort of a problem, unless say I were to actually model the drift events on a computer and "manually" calculate probabilities given a certain set of assumptions. Do i trust biostatisticians to get this right? No. (Yes, I've worked with them)