Relaxed neighbor joining: A fast distance-based phylogenetic tree construction method
Relaxed neighbor joining: A fast distance-based phylogenetic tree construction method
复制标题
DOI：
10.1007/s00239-005-0176-2
复制
发表时间：
2006-06-01
影响因子：
3.9
通讯作者：
Foster, J
中科院分区：
文献类型：
--
作者：
Evans, J;Sheneman, L;Foster, J
随着大量的序列数据变得容易获得，我们构建非常大的系统发育树的能力变得更加重要。邻居连接(Neighbor Join，NJ)是一种广泛使用的基于距离的系统发育树构建方法，历史上一直被认为是快速的，但对于从越来越大的数据集构建树来说，它的速度太慢了。我们开发了NJ的一个快速变体，称为松弛邻居加入(RNJ)，并进行了实验来衡量比NJ的速度提高。由于RNJ算法的重复运行生成了重复NJ运行生成的树的超集，因此我们还评估了树的质量。RNJ比NJ快得多，两种算法生成的树的质量非常相似。结果表明，RNJ是NJ的合理替代品，特别适用于涉及大量分类群或高度重复的程序，如自举。
Our ability to construct very large phylogenetic trees is becoming more important as vast amounts of sequence data are becoming readily available. Neighbor joining (NJ) is a widely used distance-based phylogenetic tree construction method that has historically been considered fast, but it is prohibitively slow for building trees from increasingly large datasets. We developed a fast variant of NJ called relaxed neighbor joining (RNJ) and performed experiments to measure the speed improvement over NJ. Since repeated runs of the RNJ algorithm generate a superset of the trees that repeated NJ runs generate, we also assessed tree quality. RNJ is dramatically faster than NJ, and the quality of resulting trees is very similar for the two algorithms. The results indicate that RNJ is a reasonable alternative to NJ and that it is especially well suited for uses that involve large numbers of taxa or highly repetitive procedures such as bootstrapping.