Eigen M, Winkler-Oswatitsch R, Dress A
Max-Planck-Institut für Biophysikalische Chemie, Göttingen, Federal Republic of Germany.
Proc Natl Acad Sci U S A. 1988 Aug;85(16):5913-7. doi: 10.1073/pnas.85.16.5913.
A statistical method of comparative sequence analysis that combines horizontal and vertical correlations among aligned sequences is introduced. It is based on the analysis mainly of quartet combinations of sequences considered as geometric configurations in sequence space. Numerical invariants related to relative internal segment lengths are assigned to each such configuration and statistical averages of these invariants are established. They are used for internal calibration of the topology of divergence and for quantitative determination of the noise level. Comparison of computer simulations with experimental data reveals the high sensitivity of assignment of basic topologies even if much randomized. In addition, these procedures are checked by vertical analysis of the aligned sequences to allow the study of divergences with positionally varying substitution probabilities.
介绍了一种结合比对序列间水平和垂直相关性的比较序列分析统计方法。它主要基于对序列四重组合的分析,这些序列四重组合在序列空间中被视为几何构型。与相对内部片段长度相关的数值不变量被分配给每个这样的构型,并建立这些不变量的统计平均值。它们用于分歧拓扑的内部校准和噪声水平的定量测定。计算机模拟与实验数据的比较表明,即使存在大量随机化,基本拓扑的分配仍具有很高的灵敏度。此外,通过对比对序列的垂直分析来检验这些程序,以便研究具有位置变化替换概率的分歧。