Tamura Koichiro, Kumar Sudhir
Department of Biological Sciences, Tokyo Metropolitan University, Tokyo, Japan.
Mol Biol Evol. 2002 Oct;19(10):1727-36. doi: 10.1093/oxfordjournals.molbev.a003995.
Most of the sophisticated methods to estimate evolutionary divergence between DNA sequences assume that the two sequences have evolved with the same pattern of nucleotide substitution after their divergence from their most recent common ancestor (homogeneity assumption). If this assumption is violated, the evolutionary distance estimated will be biased, which may result in biased estimates of divergence times and substitution rates, and may lead to erroneous branching patterns in the inferred phylogenies. Here we present a simple modification for existing distance estimation methods to relax the assumption of the substitution pattern homogeneity among lineages when analyzing DNA and protein sequences. Results from computer simulations and empirical data analyses for human and mouse genes are presented to demonstrate that the proposed modification reduces the estimation bias considerably and that the modified method performs much better than the LogDet methods, which do not require the homogeneity assumption in estimating the number of substitutions per site. We also discuss the relationship of the substitution and mutation rate estimates when the substitution pattern is not the same in the lineages leading to the two sequences compared.
大多数用于估计DNA序列间进化差异的复杂方法都假定，这两条序列自最近共同祖先分歧后，是以相同的核苷酸替换模式进化的（齐性假设）。如果违背了这一假设，所估计的进化距离将会有偏差，这可能导致分歧时间和替换率的估计出现偏差，并可能在推断的系统发育中产生错误的分支模式。在此，我们针对现有距离估计方法提出一种简单的修正，以便在分析DNA和蛋白质序列时放宽谱系间替换模式齐性的假设。给出了针对人类和小鼠基因的计算机模拟及实证数据分析结果，以证明所提出的修正能大幅降低估计偏差，且修正后的方法比LogDet方法表现要好得多，后者在估计每个位点的替换数时不需要齐性假设。我们还讨论了在导致所比较的两条序列的谱系中替换模式不同时，替换率和突变率估计值之间的关系。
