Potential pitfalls of modelling ribosomal RNA data in phylogenetic tree reconstruction: evidence from case studies in the Metazoa.
Potential pitfalls of modelling ribosomal RNA data in phylogenetic tree reconstruction: evidence from case studies in the Metazoa.
复制标题
DOI：
10.1186/1471-2148-11-146
复制
发表时间：
2011-05-27
影响因子：
3.4
通讯作者：
Kjer KM
中科院分区：
文献类型：
--
作者：
Letsch HO;Kjer KM
未能解释核糖体RNA （rRNA）基因螺旋区域的共变模式有可能误导对数据的系统发育信号的估计。此外，类群之间的极端长度变化，加上区域替代率的变化，可能会误导rRNA序列的比对，从而扭曲随后的树重建。然而，系统发育方法学的最新发展现在允许在基于rRNA序列的比对和树重建分析中全面整合二级结构，这已被证明可以纠正这些问题。在这里，我们探索了RNA替代模型的潜力，以及特定模型设置与rRNA茎中固有的共变异模式和环区之间的替代率变化之间的相互作用。我们发现了RNA替代模型对树重建分析的明确影响。特定RNA模型在树木重建中的应用受到了树干区域共变位点的适当建模和某些环区过度同质性之间的相互作用的阻碍。当单链区域过度同塑时，RNA模型往往无法恢复合理的树，因为当共变位点基本上被降低权重时，这些区域贡献了更大比例的数据。在这种情况下，RNA6A模型优于所有其他模型，包括参数化程度更高的RNA7和RNA16模型。我们的结果描述了在螺旋区域的相互依赖关系的估计增加的准确性与缺乏系统发育信号的放大位置的风险之间的权衡。因此，我们可以得出结论，在应用rRNA共变模型时需要谨慎，并建议对环路区域进行独立的系统发育信号筛选，并在它们与随机噪声无法区分时将其消除。除了共变和同质性外，其他因素，如取代率的非平稳性和碱基组成的异质性，也会破坏核糖体RNA数据的信号。所有这些因素决定了对rRNA数据中进化模式的复杂估计，就像其他分子数据需要同样复杂（但不同）的校正一样。
Failure to account for covariation patterns in helical regions of ribosomal RNA (rRNA) genes has the potential to misdirect the estimation of the phylogenetic signal of the data. Furthermore, the extremes of length variation among taxa, combined with regional substitution rate variation can mislead the alignment of rRNA sequences and thus distort subsequent tree reconstructions. However, recent developments in phylogenetic methodology now allow a comprehensive integration of secondary structures in alignment and tree reconstruction analyses based on rRNA sequences, which has been shown to correct some of these problems. Here, we explore the potentials of RNA substitution models and the interactions of specific model setups with the inherent pattern of covariation in rRNA stems and substitution rate variation among loop regions. We found an explicit impact of RNA substitution models on tree reconstruction analyses. The application of specific RNA models in tree reconstructions is hampered by interaction between the appropriate modelling of covarying sites in stem regions, and excessive homoplasy in some loop regions. RNA models often failed to recover reasonable trees when single-stranded regions are excessively homoplastic, because these regions contribute a greater proportion of the data when covarying sites are essentially downweighted. In this context, the RNA6A model outperformed all other models, including the more parametrized RNA7 and RNA16 models. Our results depict a trade-off between increased accuracy in estimation of interdependencies in helical regions with the risk of magnifying positions lacking phylogenetic signal. We can therefore conclude that caution is warranted when applying rRNA covariation models, and suggest that loop regions be independently screened for phylogenetic signal, and eliminated when they are indistinguishable from random noise. In addition to covariation and homoplasy, other factors, like non-stationarity of substitution rates and base compositional heterogeneity, can disrupt the signal of ribosomal RNA data. All these factors dictate sophisticated estimation of evolutionary pattern in rRNA data, just as other molecular data require similarly complicated (but different) corrections.
登录
查看更多内容
影响因子：
4.1
作者：
Hudelot, C;Gowri-Shankar, V;Higgs, PG
通讯作者：
Higgs, PG
影响因子：
10.7
作者：
Gibson, A;Gowri-Shankar, V;Rattray, M
通讯作者：
Rattray, M
影响因子：
5.5
作者：
Keller A;Förster F;Müller T;Dandekar T;Schultz J;Wolf M
通讯作者：
Wolf M
影响因子：
3.9
作者：
Erpenbeck, Dirk;Nichols, Scott A.;Woerheide, Gert
通讯作者：
Woerheide, Gert
DOI：
10.1016/0167-4781(94)90028-0
发表时间：
1994-03-01
期刊：
BIOCHIMICA ET BIOPHYSICA ACTA-GENE STRUCTURE AND EXPRESSION
影响因子：
--
作者：
KJER, KM;BALDRIDGE, GD;FALLON, AM
通讯作者：
FALLON, AM