The study of evolution is an essential task in predicting the variability of species, especially for pathogens such as viruses. One of the main stages of evolutionary analysis is constructing a phylogenetic tree.
This work is devoted to developing a new approach for visualization of the phylogenetic tree, which is based on reconstructing the evolutionary trajectory of a taxon in three-dimensional space. An evolutionary trajectory is a path that connects a particular taxon and the root of the tree. By reconstructing ancestral sequences and applying one-hot-encoding, each tree node is represented as a multidimensional object, then mapped into three-dimensional space using the embedding method, due to which, evolutionary paths from leaves to the tree’s root are generated. This approach makes it possible to visualize rapid changes in evolutionary direction, both locally and globally. The results are based on the experiments on visualization of the evolutionary trajectory of the H3N2 influenza virus and the development of a publicly available web platform called PhyloTraVis. They suggest the application of our approach for early detection of changes in the direction of evolution, the study of evolutionary dynamics, evaluation of emerging novel virus variants, and modeling of possible antigenic diversity, which are important tasks in computational virology.