Jackknife and Bootstrap Tests of the Composition Vector Trees

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

Thus our justification of CVTrees goes in two steps. First, we show that the CVTrees based on whole proteomes (to be called the original CVTrees hereafter) are robust, and the branchings agree well with the corresponding taxonomy. This has been done in our previous studies (1, (5) (6) (7) (8) (9) . Second, in this paper, trees obtained by bootstrap or jackknife re-sampling of the datasets are compared to the original CVTrees at the same K in terms of topological distances between them. As CVTrees are calculated by Neighbor-Joining algorithm (11) , which is a robust and quartet-based method (12) , our strategy reminds what adopted by Rzhetsky and Nei (13) in estimating minimal-evolution trees by comparison with the corresponding NJ-tree. We make emphasis on topology, i.e., branching schemes, of the trees as they are directly related to taxonomy.