Resolving difficult phylogenetic questions: why more sequences are not enough.

Philippe H, Brinkmann H, Lavrov DV, Littlewood DT, Manuel M, Wörheide G, Baurain D. · 2011

Read Resolving difficult phylogenetic questions: why more sequences are not enough. by Philippe H, Brinkmann H, Lavrov DV, Littlewood DT, Manuel M, Wörheide G, Baurain D. (2011) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc3057953.txt.

Introduction

Philippe and coauthors argued in 2011 that simply gathering more DNA sequences will not settle the hardest questions about the tree of life. Their essay responds to genome scale studies of early animal evolution that reached strongly conflicting trees despite huge datasets. Each rival tree tells a very different story about how simple or complex the first animals were.

The trouble, they argue, is not a lack of data but systematic error inside it. Fast evolving lineages, poor models, and contaminated or misaligned sequences can pull a tree the wrong way with great statistical confidence. More data then makes a wrong answer look more certain, not less. The authors urge better models, careful data selection, and slower thinking over brute volume. Their piece is a caution to a field dazzled by sequencing speed.

Catalog record

CREATORPhilippe H, Brinkmann H, Lavrov DV, Littlewood DT, Manuel M, Wörheide G, Baurain D.
YEAR2011
COLLECTIONPAPERS
FORMATTXT
FILE SIZE58 KB
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC3057953

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Resolving Difficult Phylogenetic Questions: Why More Sequences Are Not Enough By Hervé Philippe, Henner Brinkmann, Dennis V Lavrov, D Timothy J Littlewood, Michael Manuel, Gert Wörheide, Denis Baurain, David Penny Published in: PLoS Biology DOI: 10.1371/journal.pbio.1000602 In the quest to reconstruct the Tree of Life, researchers have increasingly turned to phylogenomics, the inference of phylogenetic relationships using genome-scale data (Box 1). Mesmerized by the sustained increase in sequencing throughput, many phylogeneticists entertained the hope that the incongruence frequently observed in studies using single or a few genes [1] would come to an end with the generation of large multigene datasets. Yet, as so often happens, reality has turned out to be far more complex, as three recent large-scale analyses, one published in PLoS Biology [2]–[4], make clear. The studies, which deal with the early diversification of animals, produced highly incongruent (Box 2) findings despite the use of considerable sequence data (see Figure 1). Clearly, merely adding more sequences is not enough to resolve the inconsistencies. Box 1. From Phylogenetics to PhylogenomicsPhylogenetics, the determination of evolutionary relationships among organisms, is central to our understanding of the evolution of life. For instance, the three phylogenies of Figure 1 entail profoundly different interpretations about the complexity of the common ancestor of all animals. Important body plan characters (e.g., neurosensory and digestive systems and muscle cells) are found in cnidarians, ctenophores, and bilaterians but not in sponges and placozoans. According to the phylogenies of Schierwater et al. [4] and Dunn et al. [2], the taxonomic distribution of these characters implies either (i) that the ancestral metazoan already featured these traits and that sponges (and placozoans) have secondarily lost them or (ii) that these characters were acquired several times independently by convergence (e.g., in the cnidarian + ctenophore and in the bilaterian lineages, according to the tree of Figure 1A). In contrast, the phylogeny of Philippe et al. [3] is more congruent with morphological characters and compatible with a simple metazoan ancestor and a later emergence of these characters only once, in the lineage leading to the common ancestor of coelenterates (cnidarians+ctenophores) and bilaterians.Phylogenies are generally depicted as trees (which are non-reticulated graphs, as in Figure 1) because vertical evolution is undisputedly the primary mechanism of inheritance for genetic material. However, the existence of horizontal transmission (e.g., hybridization of closely related taxa, organelle acquisition through endosymbiosis and horizontal gene transfer) makes phylogenetic trees only pragmatic approximations, which will probably be replaced by phylogenetic networks in the long term (particularly for unicellular organisms).Recently, phylogenomics, the use of genomic data to infer evolutionary relationships, has emerged as a new domain of phylogenetics. The main strength of phylogenomics is the drastic reduction in random (or sampling) error brought by the use of large (multigene) datasets. Numerous approaches can be used to take advantage of genomic data (for review see [49]). Briefly, new methods based on oligonucleotide content, gene content, or intron positions look promising (as shown by their ability to yield reasonable trees) but require additional theoretical developments to achieve their full potential. That is why the two most popular phylogenomic approaches are simple extensions of the standard phylogenetics methods applied to single-gene datasets. The first, known as the “supermatrix” (or superalignment), consists in concatenating numerous orthologous genes into a single supergene, which is analyzed using standard methods (or slightly modified methods such as separate models allowing for multiple sets of branch lengths [50]). The second, “supertree,” approach takes the opposite path by first inferring a tree for each gene in the dataset and then combining these individual trees into a single supertree. The supermatrix approach is the most commonly used, in agreement with the handful of studies suggesting that it offers greater accuracy than the supertree [13],[51], though this remains to be formally demonstrated. Box 2. Glossary Homology/orthology/paralogy/xenology: Genes that derive from a common ancestor are termed homologs. Two homologous genes are orthologous if they diverged through a speciation event. In contrast, paralogs originate by duplication of a single gene within a given lineage, whereas xenologs result from the horizontal transfer of a gene from a donor species to a receiver species (which might eventually get its original copy replaced by the xenolog). Homoplasy/convergence: Spurious similarity due to convergence or reversion and not to common ancestry is termed homoplasy. Convergence describes the independent acquisition by separate evolutionary lineages of the same nucleotide (or amino acid) at a given position. This is a direct consequence of multiple substitutions. Incomplete lineage sorting: The transient retention of ancestral polymorphisms across speciation events. Speciations compressed in time and large reproductive populations both increase the likelihood of this phenomenon. Considering three lineages having rapidly diverged, by chance some sequence positions will be shared between one pair, while others will be shared between another pair, and yet others between the third possible pair, hence blurring the phylogenetic signal on the corresponding branches. Incongruence: Two (or more) phylogenetic trees are said to be incongruent when they exhibit conflicting branching orders (i.e., topologies) and cannot be superimposed. This implies that at least one node (also known as a bipartition) present in one tree is not found in the other(s), where it is replaced by alternative groupings of taxa. Model of sequence evolution: A statistical description of the process of substitution in nucleotide or amino acid sequences. Complex models better approximate the evolutionary process but at the expense of more parameters and computational time. As parameter-rich models require more data to behave properly, they have become really useful with the advent of phylogenomic datasets. Monophyly: To be considered monophyletic, a taxonomic group must satisfy two conditions: (i) all its taxa must derive from a single ancestor and, reciprocally, (ii) all taxa deriving from this common ancestor must belong to the group. Non-phylogenetic signal: The combination of different kinds of structured noise (e.g., undetected homoplasies) that compete with the genuine phylogenetic signal during tree reconstruction. Even if the non-phylogenetic content is partly a property of a multiple sequence alignment (notably related to its saturation level), the non-phylogenetic signal actually inferred heavily depends on the method and the model of evolution selected. In probabilistic methods, the non-phylogenetic signal mainly results from the data violating the model of sequence evolution. These violations arise because our models are inevitably oversimplified in comparison to the complexity of the natural evolutionary process. Eventually, the apparent signal analyzed will be a blend of phylogenetic and non-phylogenetic signal. Outgroup/ingroup: Nearly all tree reconstruction methods produce unrooted trees, in which inferred relationships do not convey any information about the direction of time. To root a tree and turn it into a phylogeny, one has to include in the analysis a group of taxa that are known to be outside the group under study. This reference group is termed the outgroup, while the taxa of interest make the ingroup. Patristic distance: The sum of the lengths of the branches that connect two nodes in a phylogenetic tree, where those nodes are typically terminal nodes representing extant taxa. It is thus an inferred distance (taking into account multiple substitutions) greater than the uncorrected distance directly computed from the number of differences observed between the two corresponding sequences in the alignment. Phylogenetic signal/synapomorphy: The substitutions occurring along a

Read the complete text