Split-inducing indels in phylogenomic analysis

Alexander Donath,Peter F Stadler

doi:10.1186/s13015-018-0130-7

Abstract

BackgroundMost phylogenetic studies using molecular data treat gaps in multiple sequence alignments as missing data or even completely exclude alignment columns that contain gaps.ResultsHere we show that gap patterns in large-scale, genome-wide alignments are themselves phylogenetically informative and can be used to infer reliable phylogenies provided the gap data are properly filtered to reduce noise introduced by the alignment method. We introduce here the notion of split-inducing indels (splids) that define an approximate bipartition of the taxon set. We show both in simulated data and in case studies on real-life data that splids can be efficiently extracted from phylogenomic data sets.ConclusionsSuitably processed gap patterns extracted from genome-wide alignment provide a surprisingly clear phylogenetic signal and an allow the inference of accurate phylogenetic trees.

Highlights

Most phylogenetic studies using molecular data treat gaps in multiple sequence alignments as missing data or even completely exclude alignment columns that contain gaps
Gaps in multiple sequence alignments are usually seen as a nuisance in molecular phylogenetics
Recent studies have indicated that biases may be introduced when indels are included without precautions in Bayesian and Maximum Likelihood phylogenies [9, 10]

Summary

Introduction

Most phylogenetic studies using molecular data treat gaps in multiple sequence alignments as missing data or even completely exclude alignment columns that contain gaps. Gaps in multiple sequence alignments are usually seen as a nuisance in molecular phylogenetics. Gaps are treated as missing data or alignment columns with gaps are even removed completely. Stochastic models of sequence evolution that deal explicitly with insertions and deletions (indels) have been investigated only recently [1, 2]. The inclusion of insertion and deletion (indel) characters proved useful in the analysis of the phylogeny of the Arctoidea (Mammalia: Carnivora) [6], neognathous birds [7], or fungal families [8]. Recent studies have indicated that biases may be introduced when indels are included without precautions in Bayesian and Maximum Likelihood phylogenies [9, 10]

Methods

Results

Conclusion