QNet: An Agglomerative Method for the Construction of Phylogenetic Networks from Weighted Quartets

We present QNet, a method for constructing split networks from weighted quartet trees. QNet can be viewed as a quartet analogue of the distance-based Neighbor-Net (NNet) method for network construction. Just as NNet, QNet works by agglomeratively computing a collection of circular weighted splits of the taxa set which is subsequently represented by a planar split network. To illustrate the applicability of QNet, we apply it to a previously published Salmonella data set. We conclude that QNet can provide a useful alternative to NNet if distance data are not available or a character-based approach is preferred. Moreover, it can be used as an aid for determining when a quartet-based tree-building method may or may not be appropriate for a given data set. QNet is freely available for download.