Inferring metric trees from weighted quartets via an intertaxon distance (Q786063)

An ametric phylogenetic tree associated with a set of taxa induces weighted root triplets and weighted quartets for all subsets of three and four taxa, respectively. New extensions of the topological quartet and root triplets are developed in the article. Since the weights in merging units of root triplets and quartets can be inferred from the topological gene tree data within the MSC, it is possible to estimate these new intertaxon distances in the species tree only from the topological gene trees. While the ability to derive a metric tree of species is similar to that of STEAC, the authors' approach does not use information about the metric gene tree, and thus its consistency does not depend on any assumptions about the relationship of metric units on the gene trees and the species tree. Considering a rooted metric tree, a repeated tree metric is introduced, and the internal lengths of the edges become the product of their original lengths and an integer factor that depends on the location of the edge in the topological tree. Although this technique is known, its value lies in allowing an algorithmic approach to derive a score tree from its weighted root triplets, even in the presence of noise. For a non-rooted scorecard, was defined a repeated metric similar to using weighted quartets. Although the algorithms developed by the authors provide statistically consistent estimates of the species tree based on gene trees within the MSC model, several factors will influence their practical performance. First, even if gene trees are selected from the MSC without error, the algorithm cannot always be expected to infer the underlying species tree from a finite sample of gene trees. Second, if the input gene trees for the algorithm are derived from sequences that have been modeled along the gene trees within some standard replacement model, it is possible to expect an inference error in the gene trees due to the finite length of the sequences. Since neither the MSC nor the substitution models can accurately describe the true processes, there is an additional error due to incorrect model specification. Although the effectiveness of phylogenetic inference methods when the model is incorrectly specified is rarely studied, modeling can provide insight into the consequences of the first two problems.