Tree Construction

Methods for inference of evolutionary trees typically follow one of two approaches: (1) search, and (2) construction. In a search method, many possible trees are tested by traversing a space of possibilities in a systematic way. An inferred tree or set of trees is chosen from among the best trees examined in the search space.

In a construction method, however, the inferred tree is not chosen but is instead built from features observed in the data. Rather than choosing the inferred tree from among a set of possible trees, inference proceeds by constructing a tree based on the data, with features specified by a systematic procedure. Agglomeration methods such as neighbor-joining or UPGMA are construction methods in that they sequentially add leaves and internal nodes in building a tree from data.

The carving of this tree-based work of art, “The Stanford Legacy” by Don Yeomans, can be viewed as having followed an analogous procedure to tree construction methods such as neighbor-joining or UPGMA. The totem pole provides a tree-based representation of the university’s founding by Jane and Leland Stanford after the death of Leland Stanford Jr. It includes a child figure with angel wings at the top, and child figures climbing the tears of a weeping woman. In the analogy to tree construction, the “data” for the artwork are the facts of the university’s founding, and the artist’s “algorithm” for “inferring” the artwork from the “data” is analogous to the application of the neighbor-joining algorithm on a genetic data set to construct an inferred tree.

Photo: Noah Rosenberg, January 17, 2020