On the impact of uncertain gene tree rooting on duplication-transfer-loss reconciliation

Abstract

BackgroundDuplication-Transfer-Loss (DTL) reconciliation is a powerful and increasingly popular technique for studying the evolution of microbial gene families. DTL reconciliation requires the use of rooted gene trees to perform the reconciliation with the species tree, and the standard technique for rooting gene trees is to assign a root that results in the minimum reconciliation cost across all rootings of that gene tree. However, even though it is well understood that many gene trees have multiple optimal roots, only a single optimal root is randomly chosen to create the rooted gene tree and perform the reconciliation. This remains an important overlooked and unaddressed problem in DTL reconciliation, leading to incorrect evolutionary inferences. In this work, we perform an in-depth analysis of the impact of uncertain gene tree rooting on the computed DTL reconciliation and provide the first computational tools to quantify and negate the impact of gene tree rooting uncertainty on DTL reconciliation.ResultsOur analysis of a large data set of over 4500 gene families from 100 species shows that a large fraction of gene trees have multiple optimal rootings, that these multiple roots often, but not always, appear closely clustered together in the same region of the gene tree, that many aspects of the reconciliation remain conserved across the multiple rootings, that gene tree error has a profound impact on the prevalence and structure of multiple optimal rootings, and that there are specific interesting patterns in the reconciliation of those gene trees that have multiple optimal roots.ConclusionsOur results show that unrooted gene trees can be meaningfully reconciled and high-quality evolutionary information can be obtained from them even after accounting for multiple optimal rootings. In addition, the techniques and tools introduced in this paper make it possible to systematically avoid incorrect evolutionary inferences caused by incorrect or uncertain gene tree rooting. These tools have been implemented in the phylogenetic reconciliation software package RANGER-DTL 2.0, freely available from http://compbio.engr.uconn.edu/software/RANGER-DTL/.

FastMulRFS: fast and accurate species tree estimation under generic gene duplication and loss models

“…Species trees are important models that can be used to address many biological questions, for example how is biodiversity created/maintained and how do species adapt to their environments ( Cracraft et al , 2004 ). There is also a vast literature regarding gene tree reconciliation , where gene trees are compared to an established species tree in order to understand how genes evolved (for some of the recent literature on this question, see Delabre et al , 2020 ; Dondi et al , 2019 ; El-Mabrouk and Noutahi, 2019 ; Hasić and Tannier, 2019 ; Jacox et al , 2016 ; Kundu and Bansal, 2018 ; Lai et al , 2017 ; Muhammad et al , 2018 ). However, in most cases, species trees are not known in advance and instead must be estimated.…”

Assessing the accuracy of phylogenetic rooting methods on prokaryotic gene families

“…Parsimonious Duplication-Transfer-Loss reconciliation (DTL) rooting. DTL rooting [30,31] is designed specifically for rooting prokaryotic gene trees and is based on parsimoniously reconciling an unrooted gene tree with a known rooted species tree. It accounts for some of the major evolutionary processes responsible for gene tree/species tree discord; specifically gene duplications, gene losses, and horizontal gene transfers.…”

“…These methods are designed specifically for gene tree rooting and work by reconciling an unrooted gene tree with a known rooted species tree. This reconciliation is computed based on the duplication-loss model for eukaryotic gene families [26] and on the duplication-transfer-loss (DTL) reconciliation loss model for prokaryotic gene families [27][28][29][30][31]. Essentially, these methods use the known species tree root to inform the placement of gene tree roots and rely on the fact that species tree root inference is often less complex since it tends to be estimated from more informative and stable data sets.…”

The Clade Displacement Index: how to detect horizontal gene transfers in unrooted gene trees

“…At this point, to our knowledge, there is a lack of systematic and comprehensive studies on how to properly estimate these costs and the impact of their misspecification, although some works start to appear [17, 29, 30]. Furthermore, this method often returns multiple root candidates, and is sensitive to errors in gene tree reconstruction [19, 31]. Overall, the correct placement of the root of a gene or a species tree is far from straightforward, sometimes to the point that detailed studies for single groups of organisms are needed [32, 33].…”