Kevin Atteson - Identifiability and consistency in phylogenetics

In 1978, Joe Felsenstein demonstrated that the popular parsimony method of phylogenetics is statistically inconsistent, that is, that it need not converge to the true tree as amount of data goes to , for a simple stochastic model of the mutation of genetic sequences. Since then, the consistency question has been answered for numerous methods under a variety of stochastic mutation models. When mutation rates vary along the genetic sequence according to a completely unknown distribution, Steel et. al. have shown that the phylogenetic tree is unidentifiable, that is, that there exists no consistent algorithm. The author intends to briefly review results in this area, presenting his own results in this context, namely, the consistency of neighbor-joining methods and the identifiability of the phylogenetic tree when rates are distributed according to a Gamma distribution.