Lachance Marc-André, Hurtado Emilia, Hsiang Tom
Department of Biology, University of Western Ontario, London, Ontario, Canada, N6A 5B7.
School of Environmental Sciences, University of Guelph, Guelph, Ontario, Canada, N1G 2W1.
Yeast. 2016 Jul;33(7):261-75. doi: 10.1002/yea.3163. Epub 2016 May 10.
Draft genomes of 55 strains representing all known large-spored Metschnikowia species were used to construct a robust phylogeny of these yeasts found in association with flower-visiting insects. The genomes were annotated with reference to Clavispora lusitaniae. From 3016 orthologues identified, 1061 were present in all strains with enough overlap to generate alignments of 500 bp or more. We constructed trees for all those alignments and evaluated their accuracy from their ability to resolve each of 22 sets of conspecifics correctly as sister taxa. Neighbour-joining identified species membership better than maximum likelihood, as did trees based on larger gene alignments. However, correct species assignment was not predictive of a gene's ability to resolve deeper topologies, which were more reliably identified by maximum likelihood analyses of large concatenations. Specifically, 14 trees based on independent concatenations ca. 100 kb in length were topologically consistent with a tree based on a single, large concatenation (1 410 065 positions), lending a high degree of confidence to the stability of the phylogeny. A tree based on a concatenation of intergenic regions (112 136 positions) was also congruent. Again, the best predictor of phylogenetic signal quality of a gene was the size of the alignment. Bootstraps were not always good indicators of phylogenetic quality, as they were sometimes affected by clade size. A tree constructed from a presence-absence matrix of all annotated genes was remarkably congruent with sequence-based phylogenies, suggesting that gain or loss of genes is worth exploring further as a phylogenetically significant event. Copyright © 2016 John Wiley & Sons, Ltd.
利用代表所有已知大孢子梅奇酵母属物种的55个菌株的基因组草图，构建了这些与访花昆虫相关的酵母的稳健系统发育树。这些基因组参考葡萄牙棒孢酵母进行注释。从鉴定出的3016个直系同源物中，有1061个存在于所有菌株中，且有足够的重叠以生成500bp或更长的比对。我们为所有这些比对构建了树，并根据它们将22组同种菌株正确解析为姐妹类群的能力来评估其准确性。邻接法比最大似然法能更好地鉴定物种归属，基于更大基因比对构建的树也是如此。然而，正确的物种归属并不能预测一个基因解析更深层次拓扑结构的能力，通过对大的串联序列进行最大似然分析能更可靠地鉴定这些拓扑结构。具体而言，基于约100kb长度的独立串联序列构建的14棵树在拓扑结构上与基于单个大串联序列（1410065个位点）构建的树一致，这为系统发育的稳定性提供了高度的可信度。基于基因间区域串联序列（112136个位点）构建的树也与之相符。同样，一个基因的系统发育信号质量的最佳预测指标是比对的大小。自展值并不总是系统发育质量的良好指标，因为它们有时会受到分支大小的影响。由所有注释基因的存在-缺失矩阵构建的树与基于序列的系统发育树非常一致，这表明基因的获得或丢失作为一个具有系统发育意义的事件值得进一步探讨。版权所有©2016约翰威立父子有限公司。
