Jackknife and Bootstrap Tests of the Composition Vector Trees

Zuo, Guanghong; Xu, Zhao; Yu, Hongjie; Hao, Bailin

We used the topological distance to measure the difference between phylogenetic trees. The definition of topological distance could be found in the previous literature (14, 15) . An unrooted tree with N terminal leaves has N 3 internal edges. Cutting any of these internal edges defines a split of the set of leaves into two subsets. To measure the distance between two trees constructed for the same set of leaves, we compare the two lists of split-trees obtained by cutting each of the N 3 internal edges. If the two lists are identical up to reordering, the two trees have the same topology and the topological distance d T =0. In general, the topological distance is defined as (16) : d T = 2×(number of distinct split-trees) The factor 2 was introduced to incorporate more general cases of multi-furcating nodes (CV method yields only bifurcating trees). Therefore, if the two trees have entirely different topologies, the topological distance between them reaches the maximal value d T = 2×(N 3). We have written a program to implement the definition of topological distance. In fact, we adopt a relative topological distance by dividing the calculated d T by its maximal possible value 2×(N 3), thus the factor 2 drops out and the relative distance varies between 0 and 1.