Exclusion of fast evolving genes or fast evolving sites produces different archaean phylogenies.

Citation metadata

Date: May 2022
Publisher: Elsevier B.V.
Document Type: Report; Brief article
Length: 332 words

Document controls

Main content

Abstract :

Keywords Fast evolving genes; Fast evolving sites; Evolutionary rates; Archaea phylogeny; Phylogenetic robustness; Topology Highlights * Reconstructed topologies differ when fast evolving genes or fast evolving sites are removed from the dataset. * Recent Archaea phylogenies are not robust to different data manipulation approaches. * Filtering fast evolving genes or fast evolving sites support opposite mono- or paraphyletic status for Class I and Class II methanogens. * Comparison of filtering strategies is important to determine phylogenetic robustness. Abstract The large size of modern datasets has inspired a variety of strategies to alter genes, sites and/or species compositions to improve the accuracy of phylogenetic reconstructions. Each of these data-filtering approaches leads to the exclusion of different types of data, which is known to affect phylogenetic outcome. However, the choice of a filtering strategy is often subjective and the robustness of the results to alternative strategies is not usually investigated. A case study is provided by archaean phylogenies, that have recently relied on filtering fast evolving sites. Here we compare the outcomes of two filtering strategies: fast evolving genes or fast evolving sites to investigate the robustness of Archaea phylogenies to data manipulation. Our results show that the two approaches lead to different outcomes and that excluding fast evolving sites might not be as effective as removing fast evolving genes. Topologies obtained from filtering either fast evolving or random sites are twice as likely to be similar (72%) than those obtained from filtering fast genes (36%). Our results suggest that the phylogenies of Archaea from many recent studies may be driven by the data filtering choice and that robustness to different types of data manipulation should be systematically investigated. Author Affiliation: (a) Department of Biological Sciences, 118 Library dr., Oakland University, Michigan, USA (b) Center for Data Science and Big Data Analytics, 118 Library dr., Oakland University, Michigan, USA * Corresponding author. Article History: Received 25 March 2019; Revised 7 January 2022; Accepted 3 February 2022 Byline: A.A. Superson (a), F.U. Battistuzzi [battistu@oakland.edu] (a,b,*)

Source Citation

Source Citation   

Gale Document Number: GALE|A699113560