Diagonalizing the genome I: navigation in tree spaces

The orientable cover of the moduli space of real genus zero algebraic curves with marked points is a compact aspherical manifold tiled by associahedra, which resolves the singularities of the space of phylogenetic trees. The resolution maps planar metric trees to their underlying abstract representatives, collapsing and folding an explicit geometric decomposition of the moduli space into cubes. This decomposition endows the resolving space with an interesting canonical pseudometric. The second part of this paper defines a related (stacky) resolution of a space of real quadratic forms, and suggests, perhaps without much justification, that systems of oscillators parametrized by such objects may provide useful models in genomics.

💡 Research Summary

The paper tackles two intertwined problems at the interface of algebraic geometry, combinatorial topology, and computational biology. The first part studies the orientable double cover of the real moduli space M₀,n(ℝ) of genus‑zero algebraic curves with n marked points. This space, which parametrizes planar metric trees, is known to have singularities that make the classical phylogenetic tree space (the Billera–Holmes–Vogtmann, or BHV, space) non‑manifold‑like at certain “corner” points. The authors show that the orientable cover is a compact aspherical manifold that can be tiled by associahedra—high‑dimensional polytopes whose face lattice encodes the combinatorics of binary tree rotations. By explicitly decomposing each associahedron into a product of cubes, they obtain a concrete cubical subdivision of the cover.

With this subdivision in hand, they define a “resolution map” that sends a planar metric tree to its underlying abstract tree. Geometrically the map collapses certain faces of the cubes (folding) and identifies others, thereby smoothing out the singularities of the BHV space. The key by‑product is a canonical pseudometric on the resolved space: the distance between two trees is the length of the shortest edge‑path through the cubical complex, measured in the natural Euclidean metric on each cube. This distance respects tree rotations more faithfully than the original BHV metric and is computationally tractable because it reduces to shortest‑path problems on a finite graph.

The second part of the paper shifts focus to the space of real quadratic forms. By interpreting the eigenvalue spectrum of a positive‑definite quadratic form as a labeling of the leaves of a binary tree, the authors construct a stacky (orbifold‑like) resolution analogous to the associahedral tiling. They then propose a physical interpretation: each quadratic form corresponds to a network of coupled harmonic oscillators, where eigenvalues become natural frequencies and the tree’s internal edge lengths become coupling constants. In this picture, the dynamics of the oscillator network encode the geometry of the tree space, suggesting a novel way to model high‑dimensional genomic data (e.g., gene‑expression patterns, mutational landscapes) as the state space of a dynamical system.

Although the biological application is speculative, the paper outlines concrete research directions: (1) integrating the new pseudometric into phylogenetic inference pipelines to improve tree optimization; (2) testing oscillator‑based models on real genomic datasets to see whether dynamical signatures correlate with known evolutionary relationships; (3) exploring connections between associahedral tilings and other combinatorial polytopes such as permutohedra, which may yield alternative resolutions of tree spaces.

In summary, the work provides (i) a rigorous geometric resolution of the singular BHV tree space via an associahedron‑tiled manifold, (ii) a natural, computationally friendly pseudometric on this resolved space, and (iii) a bold, physics‑inspired framework that maps quadratic‑form geometry to oscillator dynamics, opening a potential pathway for applying sophisticated topological tools to problems in genomics.