Original title: Phylogenetic least squares estimation without genetic distances
Authors: Peter B. Chi, Volodymyr M. Minin
In their article, the researchers introduce a groundbreaking method for phylogenetic estimation. Traditionally, estimating phylogenies involves multiple steps, often starting with a distance matrix, which isn’t always accurate. This new approach bypasses this initial step by creating a specialized loss function directly from sequence data, eliminating the need for a pre-estimated distance matrix. Using various substitution models, they demonstrate that this method not only improves phylogenetic inference but also allows for estimating additional parameters impossible with conventional distance-based methods. Their innovation opens avenues for more precise phylogenetic inference, especially when limited data restricts traditional likelihood-based approaches, promising more accurate tree-building from specific nucleotide substitutions like synonymous or non-synonymous changes.
Original article: https://arxiv.org/abs/2311.12717