Towards a Data Reduction for the Minimum Flip Supertree Problem

TL;DR

This paper introduces a polynomial-time data reduction technique for the NP-complete Minimum Flip Supertree problem in phylogenetics, enabling preprocessing and potential solution simplification despite practical hardness.

Contribution

It presents a parameterized, polynomial-time data reduction method that preserves solution equivalence and extends to parameter-independent bounds, aiding in solving complex instances.

Findings

Data reduction preserves solution existence.

Reduction's effectiveness diminishes with instance complexity.

Practical solving remains challenging for large instances.

Abstract

In computational phylogenetics, the problem of constructing a supertree of a given set of rooted input trees can be formalized in different ways, to cope with contradictory information in the input. We consider the Minimum Flip Supertree problem, where the input trees are transformed into a 0/1/?-matrix, such that each row represents a taxon, and each column represents an inner node of one of the input trees. Our goal is to find a perfect phylogeny for the input matrix requiring a minimum number of 0/1-flips, that is, corrections of 0/1-entries in the matrix. The problem is known to be NP-complete. Here, we present a parameterized data reduction with polynomial running time. The data reduction guarantees that the reduced instance has a solution if and only if the original instance has a solution. We then make our data reduction parameter-independent by using upper bounds. This allows us to preprocess an instance, and to solve the reduced instance with an arbitrary method. Different from an existing data reduction for the consensus tree problem, our reduction allows us to draw conclusions about certain entries in the matrix. We have implemented and evaluated our data reduction. Unfortunately, we find that the Minimum Flip Supertree problem is also hard in practice: The amount of information that can be derived during data reduction diminishes as instances get more "complicated", and running times for "complicated" instances quickly become prohibitive. Still, our method offers another route of attack for this relevant phylogenetic problem.