But at this point everyone is misinterpreting likely stochastic drift effects, sampling bias.
1) experimental evidence of mutation function. Ideally animal transmission model.
2) evidence of prevalance shift from competition. Assuming null distribution is equal, varied transmission in controlled host-limiting environment.
The work comes from an excellent research group in HIV work. This bioinformatic pipeline will have real utility. They got a bit over their skis on title, conclusions. They deserve a strong and good-faith peer review.
This article and 'discourse' is not that.