Drawing explicit phylogenetic networks and their integration into SplitsTree

Abstract

Background: SplitsTree provides a framework for the calculation of phylogenetic trees and networks. It contains a wide variety of methods for the import/export, calculation and visualization of phylogenetic information. The software is developed in Java and implements a command line tool as well as a graphical user interface.

Results: In this article, we present solutions to two important problems in the field of phylogenetic networks. The first problem is the visualization of explicit phylogenetic networks. To solve this, we present a modified version of the equal angle algorithm that naturally integrates reticulations into the layout process and thus leads to an appealing visualization of these networks. The second problem is the availability of explicit phylogenetic network methods for the general user. To advance the usage of explicit phylogenetic networks by biologists further, we present an extension to the SplitsTree framework that integrates these networks. By addressing these two problems, SplitsTree is among the first programs that incorporates implicit and explicit network methods together with standard phylogenetic tree methods in a graphical user interface environment.

Conclusion: In this article, we presented an extension of SplitsTree 4 that incorporates explicit phylogenetic networks. The extension provides a set of core classes to handle explicit phylogenetic networks and a visualization of these networks.

Figure 1

Example of the layout optimization. The figure on the left side shows an explicit phylogenetic network. The reticulation edges of the network are shown as dashed lines. Removing the reticulation edges and integrating the auxiliary edges into the network leads to a tree structure, as shown on the right-hand side of the figure (auxiliary edges are drawn as dashed lines). The set of easily reachable edges R_v of the node v contains the reticulations r1, r2, r3, r4 and r*. The set $R_v^D$ only contains r* and is equal to ${\overline{R}}_v^D$. The placement of r* has cost 1 since the reticulation edges only cross the edge that leads to leaf g. The placement of r4 has cost 2, since the reticulation edges to the right crosses r1 and r3.

Figure 2

Example of the drawing algorithm. The figure shows the drawing of a reticulation network that we recently published [20] which is based on three gene trees described in [28].

Figure 3

Reticulate Nexus Block Schematic. Shown is a schematic of the Reticulate nexus block as it is implemented in SplitsTree. The block is divided into three parts: the Dimensions contains all information about the dimensions of the reticulate network; the Format is an optional element that describes the configuration of the reticulate network; and the TreeComponents, NettedComponents and RootComponents contain the string representation of the reticulate network.