A faster algorithm for gene-duplication problem based on rSPR local search

Abstract

The gene-duplication problem is to infer a species tree from a collection of gene trees. As we know, this problem is NP-hard and thus requires efficient heuristics. Existing heuristics for this problem performs a stepwise search of the tree space which consists of all potential species trees. Each step in heuristics is guided by an exact solution to an instance of local search problem. The local search problem is to find an optimal species tree in the tree space. Computing the mapping for gene trees and a species tree is a necessary step in the local search problem. In this paper, we give an improved algorithm to determine the mapping between gene trees and a species tree in the gene-duplication problem based on the rooted subtree pruning and regrafting (rSPR) operation. This algorithm eliminates some redundant calculations and is faster than former algorithms.