Phylogenetic Inference under the Balanced Minimum Evolution Criterion via Semidefinite Programming

TL;DR

This paper introduces a novel approach using Semidefinite Programming to improve phylogenetic inference under the Balanced Minimum Evolution criterion, demonstrating promising results on simulated and real data.

Contribution

It proposes a new SDP-based algorithm with a rounding scheme for phylogenetic tree reconstruction, a method not previously explored in this context.

Findings

Accurate phylogenetic trees reconstructed from simulated datasets.

Effective application to empirical biological datasets.

Potential for extension to other phylogenetic problems.

Abstract

In this study, we investigate the application of Semidefinite Programming (SDP) to phylogenetics. SDP is a powerful optimization framework that seeks to optimize a linear objective function over the cone of positive semidefinite matrices. As a convex optimization problem, SDP generalizes linear programming and provides tight relaxations for many combinatorial optimization problems. However, despite its many applications, SDP remains largely unused in computational biology. We argue that SDP relaxations are particularly well suited for phylogenetic inference. As a proof of concept, we focus on the Balanced Minimum Evolution (BME) problem, a widely used model in distance-based phylogenetics. We propose an algorithm combining an SDP relaxation with a rounding scheme that iteratively converts relaxed solutions into valid tree topologies. Experiments on simulated and empirical datasets show that the method enables accurate phylogenetic reconstruction. The approach is sufficiently general to be extendable to other phylogenetic problems.