[Paper Review] A structure theorem for tree-based phylogenetic networks

Momoko Hayamizu|arXiv (Cornell University)|Nov 14, 2018

Evolution and Paleontology Studies3 citations

TL;DR

This paper introduces a structure theorem for rooted binary phylogenetic networks based on maximal zig-zag trail decomposition, enabling unified and efficient solutions to key computational problems. It provides linear-time or linear-delay algorithms for deciding tree-based networks, counting, enumerating, and optimizing over subdivision trees, with extensions to non-binary networks.

ABSTRACT

Attempting to recognize a tree inside a phylogenetic network is a fundamental undertaking in evolutionary analysis. Therefore, the concept of tree-based phylogenetic networks, which was introduced by Francis and Steel, has attracted much attention of theoretical biologists in the last few years. In this context, spanning of a certain kind called trees play an essential role and there are many important computational problems about them, whose time complexity is still unclear. Against this backdrop, the present paper aims to provide a graph theoretical framework for solving different problems on subdivision in a simple and unified manner. To this end, we focus on a structure called the maximal zig-zag trail decomposition that is inherent in any rooted binary phylogenetic network $N$ and prove a structure theorem that characterizes the collection of all subdivision of $N$. Our theorem does not only imply and unify various results in the literature but also yield linear time (for enumeration, linear delay) algorithms for the following problems: given a rooted binary phylogenetic network $N$, 1) determine whether or not $N$ has a subdivision tree and find one if there exists any (decision/search problem); 2) compute the number of subdivision of $N$ (counting problem); 3) list all subdivision of $N$ (enumeration problem); and 4) find a subdivision tree to maximize or minimize a prescribed objective function (optimization problem). Importantly, the results and algorithms in this paper still hold true for some non-binary phylogenetic networks and this generalization gives a partial answer to an open question from Pons, Semple, and Steel. We also mention some statistical applications and further research directions.

Motivation & Objective

• To develop a unified graph-theoretic framework for analyzing subdivision trees in rooted binary phylogenetic networks.
• To resolve open problems regarding the time complexity of fundamental computational tasks on tree-based networks.
• To extend results to non-binary phylogenetic networks, addressing a partial answer to an open question in the literature.
• To provide efficient algorithms for decision, counting, enumeration, and optimization problems on subdivision trees.
• To support statistical applications and guide future research in evolutionary network analysis.

Proposed method

• Introduces the maximal zig-zag trail decomposition as a fundamental structural component inherent in any rooted binary phylogenetic network.
• Proves a structure theorem that characterizes the collection of all subdivision trees via this decomposition.
• Uses the decomposition to unify and generalize existing results on tree-based networks in a graph-theoretic framework.
• Designs linear-time algorithms for decision and search problems regarding the existence of subdivision trees.
• Develops linear-delay enumeration algorithms for listing all subdivision trees efficiently.
• Extends the framework to non-binary networks, maintaining algorithmic efficiency and correctness.

Experimental results

Research questions

• RQ1 How can the collection of all subdivision trees in a rooted binary phylogenetic network be systematically characterized?
• RQ2 What is the computational complexity of enumerating all subdivision trees, and can it be solved in linear delay?
• RQ3 Can efficient algorithms be developed for counting and optimizing over subdivision trees in phylogenetic networks?
• RQ4 To what extent can the framework be generalized to non-binary phylogenetic networks?
• RQ5 What structural properties of networks guarantee the existence of subdivision trees?

Key findings

• The maximal zig-zag trail decomposition provides a complete structural characterization of all subdivision trees in a rooted binary phylogenetic network.
• The structure theorem unifies and generalizes multiple prior results in the literature on tree-based networks.
• Linear-time algorithms are developed for determining the existence of a subdivision tree and finding one if it exists.
• Linear-delay enumeration algorithms are provided for listing all subdivision trees in a network.
• The framework supports efficient counting of subdivision trees and optimization over them under arbitrary objective functions.
• The results extend to non-binary phylogenetic networks, offering a partial solution to an open question posed by Pons, Semple, and Steel.