A comprehensive analysis of mammalian mitochondrial genome base composition and improved phylogenetic methods
A comprehensive analysis of mammalian mitochondrial genome base composition and improved phylogenetic methods
复制标题
DOI：
10.1093/molbev/msi012
复制
发表时间：
2005-02-01
影响因子：
10.7
通讯作者：
Rattray, M
中科院分区：
文献类型：
--
作者：
Gibson, A;Gowri-Shankar, V;Rattray, M
利用线粒体蛋白基因对哺乳动物物种进行系统发育分析在许多先前的研究中被证明是有问题的。线粒体DNA的高突变率和不同寻常的碱基组成促使我们对69种哺乳动物线粒体基因组的组成进行了详细的研究。血统之间碱基组成的大多数主要变化可归因于l链上C和T比例的变化。这些变化在所有密码子位置都是显著的，并显示影响氨基酸组成。还观察到RNA环和茎的碱基组成的相关变化。跟进之前的研究。我们研究了所有12种h链蛋白的碱基组成的变化，发现C和T比例的变化与基因组上的位置相关。基因和物种间碱基组成的差异已知会对系统发育推断方法的性能产生不利影响。因此，我们开发了一个定制的三态通用时间可逆DNA取代模型，在PHASE系统发育推断包中实现，该模型将C和T聚集成一个复合嘧啶态。我们比较了用新的三态模型得到的系统发育树和用标准的四态模型得到的系统发育树。使用三态模型的结果与最近使用大型核基因集的研究更加一致，并有助于解决使用核蛋白和线粒体蛋白的研究之间的一些明显冲突。
Phylogenetic analysis of mammalian species using mitochondrial protein genes has proved to be, problematic in many previous studies. The high mutation rate of mitochondrial DNA and unusual base, composition of several species has prompted us to conduct a detailed Study of the composition of 69 mammalian mitochondrial genomes. Most major changes in base composition between lineages can be attributed to shifts between the proportions of C and T on the L-strand. These changes are significant at all codon positions and are shown to affect amino acid composition. Correlated changes in the base composition of the RNA loops and stems are also observed. Following up from previous studies. we investigate changes in the base composition of all 12 H-strand proteins and find that variability in proportions of C and T is correlated with location on the genome. Variation in base composition across genes and species is known to adversely affect the performance of phylogenetic inference methods. We have, therefore, developed a customized three-state general time-reversible DNA substitution model, implemented in the PHASE phylogenetic inference pack-age, which lumps C and T into a composite pyrimidine state. We compare the phylogenetic tree obtained using the new three-state model with that obtained using a standard four-state model. Results using the three-state model are more, congruent with recent studies using large sets of nuclear genes and help resolve some of the apparent conflicts between studies using nuclear and mitochondrial proteins.