广义Robinson-Foulds度量
The generalized Robinson-Foulds metric
摘要：
Robinson-Foulds（RF）度量可以说是最广泛使用的系统发育树相似性度量，尽管其众所周知的缺点：例如，在树中移动单个分类单元可以导致与原始分类单元具有最大距离的树;但是如果我们去掉单一的分类单元，这两棵树是相同的。为此，我们提出了RF度量的自然扩展，其不简单地计数相同的进化枝，而是也考虑类似的进化枝。与以前的方法相比，我们的模型需要分支之间的匹配来尊重两棵树的结构，这也是经典RF度量的一个属性。不幸的是，我们证明了计算这个广义RF度量是NP难的。然后，我们提出了一个简单的非线性规划的计算，并评估它的所有对所有的比较100棵树从基准数据集。我们发现，匹配，尊重树结构显着不同，从那些不，强调这种自然条件的重要性。
英文摘要：
The Robinson-Foulds (RF) metric is arguably the most widely used measure of phylogenetic tree similarity, despite its well-known shortcomings: For example, moving a single taxon in a tree can result in a tree that has maximum distance to the original one; but the two trees are identical if we remove the single taxon. To this end, we propose a natural extension of the RF metric that does not simply count identical clades but instead, also takes similar clades into consideration. In contrast to previous approaches, our model requires the matching between clades to respect the structure of the two trees, a property that the classical RF metric exhibits, too. We show that computing this generalized RF metric is, unfortunately, NP-hard. We then present a simple Integer Linear Program for its computation, and evaluate it by an all-against-all comparison of 100 trees from a benchmark data set. We find that matchings that respect the tree structure differ significantly from those that do not, underlining the importance of this natural condition.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
