Metrics for phylogenetic networks I: generalizations of the Robinson-Foulds metric

A Fixed-Parameter Tractable Algorithm for Finding Agreement Cherry-Reduced Subnetworks in Level-1 Orchard Networks

Phylogenetic networks are increasingly being considered better suited to represent the complexity of the evolutionary relationships between species. One class of phylogenetic networks that have received a lot of attention recently is the class of orchard networks, which is composed of networks that can be reduced to a single leaf using cherry reductions. Cherry reductions, also called cherry-picking operations, remove either a leaf of a simple cherry (sibling leaves sharing a parent) or a reticulate edge of a reticulate cherry (two leaves whose parents are connected by a reticulate edge). In this article, we present a fixed-parameter tractable algorithm to solve the problem of finding a maximum agreement cherry-reduced subnetwork (MACRS) between two rooted binary level-1 networks. This is the first exact algorithm proposed to solve the MACRS problem. As proven in an earlier work, there is a direct relationship between finding an MACRS and calculating a distance based on cherry operations. As a result, the proposed algorithm also provides a distance that can be used for the comparison of level-1 networks.

Classes of explicit phylogenetic networks and their biological and mathematical significance

The evolutionary relationships among organisms have traditionally been represented using rooted phylogenetic trees. However, due to reticulate processes such as hybridization or lateral gene transfer, evolution cannot always be adequately represented by a phylogenetic tree, and rooted phylogenetic networks that describe such complex processes have been introduced as a generalization of rooted phylogenetic trees. In fact, estimating rooted phylogenetic networks from genomic sequence data and analyzing their structural properties is one of the most important tasks in contemporary phylogenetics. Over the last two decades, several subclasses of rooted phylogenetic networks (characterized by certain structural constraints) have been introduced in the literature, either to model specific biological phenomena or to enable tractable mathematical and computational analyses. In the present manuscript, we provide a thorough review of these network classes, as well as provide a biological interpretation of the structural constraints underlying these networks where possible. In addition, we discuss how imposing structural constraints on the network topology can be used to address the scalability and identifiability challenges faced in the estimation of phylogenetic networks from empirical data.

RF-Net 2: fast inference of virus reassortment and hybridization networks

MOTIVATION

A phylogenetic network is a powerful model to represent entangled evolutionary histories with both divergent (speciation) and convergent (e.g. hybridization, reassortment, recombination) evolution. The standard approach to inference of hybridization networks is to (i) reconstruct rooted gene trees and (ii) leverage gene tree discordance for network inference. Recently, we introduced a method called RF-Net for accurate inference of virus reassortment and hybridization networks from input gene trees in the presence of errors commonly found in phylogenetic trees. While RF-Net demonstrated the ability to accurately infer networks with up to four reticulations from erroneous input gene trees, its application was limited by the number of reticulations it could handle in a reasonable amount of time. This limitation is particularly restrictive in the inference of the evolutionary history of segmented RNA viruses such as influenza A virus (IAV), where reassortment is one of the major mechanisms shaping the evolution of these pathogens.

RESULTS

Here, we expand the functionality of RF-Net that makes it significantly more applicable in practice. Crucially, we introduce a fast extension to RF-Net, called Fast-RF-Net, that can handle large numbers of reticulations without sacrificing accuracy. In addition, we develop automatic stopping criteria to select the appropriate number of reticulations heuristically and implement a feature for RF-Net to output error-corrected input gene trees. We then conduct a comprehensive study of the original method and its novel extensions and confirm their efficacy in practice using extensive simulation and empirical IAV evolutionary analyses.

AVAILABILITY AND IMPLEMENTATION

RF-Net 2 is available at https://github.com/flu-crew/rf-net-2.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

The Generalized Robinson-Foulds Distance for Phylogenetic Trees

The Robinson-Foulds (RF) distance, one of the most widely used metrics for comparing phylogenetic trees, has the advantage of being intuitive, with a natural interpretation in terms of common splits, and it can be computed in linear time, but it has a very low resolution, and it may become trivial for phylogenetic trees with overlapping taxa, that is, phylogenetic trees that share some but not all of their leaf labels. In this article, we study the properties of the Generalized Robinson-Foulds (GRF) distance, a recently proposed metric for comparing any structures that can be described by multisets of multisets of labels, when applied to rooted phylogenetic trees with overlapping taxa, which are described by sets of clusters, that is, by sets of sets of labels. We show that the GRF distance has a very high resolution, it can also be computed in linear time, and it is not (uniformly) equivalent to the RF distance.

Comparative Performance of Popular Methods for Hybrid Detection using Genomic Data

Interspecific hybridization is an important evolutionary phenomenon that generates genetic variability in a population and fosters species diversity in nature. The availability of large genome scale datasets has revolutionized hybridization studies to shift from the observation of the presence or absence of hybrids to the investigation of the genomic constitution of hybrids and their genome-specific evolutionary dynamics. Although a handful of methods have been proposed in an attempt to identify hybrids, accurate detection of hybridization from genomic data remains a challenging task. In addition to methods that infer phylogenetic networks or that utilize pairwise divergence, site pattern frequency based and population genetic clustering approaches are popularly used in practice, though the performance of these methods under different hybridization scenarios has not been extensively examined. Here, we use simulated data to comparatively evaluate the performance of four tools that are commonly used to infer hybridization events: the site pattern frequency based methods HyDe and the D-statistic (i.e., the ABBA-BABA test) and the population clustering approaches structure and ADMIXTURE. We consider single hybridization scenarios that vary in the time of hybridization and the amount of incomplete lineage sorting (ILS) for different proportions of parental contributions (γ); introgressive hybridization; multiple hybridization scenarios; and a mixture of ancestral and recent hybridization scenarios. We focus on the statistical power to detect hybridization and the false discovery rate (FDR) for comparisons of the D-statistic and HyDe, and the accuracy of the estimates of γ as measured by the mean squared error for HyDe, structure, and ADMIXTURE. Both HyDe and the D-statistic are powerful for detecting hybridization in all scenarios except those with high ILS, although the D-statistic often has an unacceptably high FDR. The estimates of γ in HyDe are impressively robust and accurate whereas structure and ADMIXTURE sometimes fail to identify hybrids, particularly when the proportional parental contributions are asymmetric (i.e., when γ is close to 0). Moreover, the posterior distribution estimated using structure exhibits multimodality in many scenarios, making interpretation difficult. Our results provide guidance in selecting appropriate methods for identifying hybrid populations from genomic data.

Probabilistic Species Tree Distances: Implementing the Multispecies Coalescent to Compare Species Trees Within the Same Model-Based Framework Used to Estimate Them

Despite the ubiquitous use of statistical models for phylogenomic and population genomic inferences, this model-based rigor is rarely applied to post hoc comparison of trees. In a recent study, Garba et al. derived new methods for measuring the distance between two gene trees computed as the difference in their site pattern probability distributions. Unlike traditional metrics that compare trees solely in terms of geometry, these measures consider gene trees and associated parameters as probabilistic models that can be compared using standard information theoretic approaches. Consequently, probabilistic measures of phylogenetic tree distance can be far more informative than simply comparisons of topology and/or branch lengths alone. However, in their current form, these distance measures are not suitable for the comparison of species tree models in the presence of gene tree heterogeneity. Here, we demonstrate an approach for how the theory of Garba et al. (2018), which is based on gene tree distances, can be extended naturally to the comparison of species tree models. Multispecies coalescent (MSC) models parameterize the discrete probability distribution of gene trees conditioned upon a species tree with a particular topology and set of divergence times (in coalescent units), and thus provide a framework for measuring distances between species tree models in terms of their corresponding gene tree topology probabilities. We describe the computation of probabilistic species tree distances in the context of standard MSC models, which assume complete genetic isolation postspeciation, as well as recent theoretical extensions to the MSC in the form of network-based MSC models that relax this assumption and permit hybridization among taxa. We demonstrate these metrics using simulations and empirical species tree estimates and discuss both the benefits and limitations of these approaches. We make our species tree distance approach available as an R package called pSTDistanceR, for open use by the community.

Compression of Phylogenetic Networks and Algorithm for the Tree Containment Problem

Rooted phylogenetic networks are rooted acyclic digraphs. They are used to model complex evolution where hybridization, recombination, and other reticulation events play a role. A rigorous definition of network compression is introduced on the basis of recent studies of relationships between cluster, tree, and rooted phylogenetic networks. The concept reveals new connections between well-studied network classes, including tree-child networks and reticulation-visible networks. It also enables us to define a new class of networks for which the cluster containment problem is solvable in linear time.

Modeling Hybridization Under the Network Multispecies Coalescent

Simultaneously modeling hybridization and the multispecies coalescent is becoming increasingly common, and inference of species networks in this context is now implemented in several software packages. This article addresses some of the conceptual issues and decisions to be made in this modeling, including whether or not to use branch lengths and issues with model identifiability. This article is based on a talk given at a Spotlight Session at Evolution 2017 meeting in Portland, Oregon. This session included several talks about modeling hybridization and gene flow in the presence of incomplete lineage sorting. Other talks given at this meeting are also included in this special issue of Systematic Biology.

S-Cluster++: a fast program for solving the cluster containment problem for phylogenetic networks

Motivation

Comparative genomic studies indicate that extant genomes are more properly considered to be a fusion product of random mutations over generations (vertical evolution) and genomic material transfers between individuals of different lineages (reticulate transfer). This has motivated biologists to use phylogenetic networks and other general models to study genome evolution. Two fundamental algorithmic problems arising from verification of phylogenetic networks and from computing Robinson-Foulds distance in the space of phylogenetic networks are the tree and cluster containment problems. The former asks how to decide whether or not a phylogenetic tree is displayed in a phylogenetic network. The latter is to decide whether a subset of taxa appears as a cluster in some tree displayed in a phylogenetic network. The cluster containment problem (CCP) is also closely related to testing the infinite site model on a recombination network. Both the tree containment and CCP are NP-complete. Although the CCP was introduced a decade ago, there has been little progress in developing fast algorithms for it on arbitrary phylogenetic networks.

Results

In this work, we present a fast computer program for the CCP. This program is developed on the basis of a linear-time transformation from the small version of the CCP to the SAT problem.

Availability and implementation

The program package is available for download on http://www.math.nus.edu.sg/∼matzlx/ccp.

Displayed Trees Do Not Determine Distinguishability Under the Network Multispecies Coalescent

Recent work in estimating species relationships from gene trees has included inferring networks assuming that past hybridization has occurred between species. Probabilistic models using the multispecies coalescent can be used in this framework for likelihood-based inference of both network topologies and parameters, including branch lengths and hybridization parameters. A difficulty for such methods is that it is not always clear whether, or to what extent, networks are identifiable-that is whether there could be two distinct networks that lead to the same distribution of gene trees. For cases in which incomplete lineage sorting occurs in addition to hybridization, we demonstrate a new representation of the species network likelihood that expresses the probability distribution of the gene tree topologies as a linear combination of gene tree distributions given a set of species trees. This representation makes it clear that in some cases in which two distinct networks give the same distribution of gene trees when sampling one allele per species, the two networks can be distinguished theoretically when multiple individuals are sampled per species. This result means that network identifiability is not only a function of the trees displayed by the networks but also depends on allele sampling within species. We additionally give an example in which two networks that display exactly the same trees can be distinguished from their gene trees even when there is only one lineage sampled per species. [gene tree, hybridization, identifiability, maximum likelihood, species tree, phylogeny.].

Reconstruction of LGT networks from tri-LGT-nets

Phylogenetic networks have gained attention from the scientific community due to the evidence of the existence of evolutionary events that cannot be represented using trees. A variant of phylogenetic networks, called LGT networks, models specifically lateral gene transfer events, which cannot be properly represented with generic phylogenetic networks. In this paper we treat the problem of the reconstruction of LGT networks from substructures induced by three leaves, which we call tri-LGT-nets. We first restrict ourselves to a class of LGT networks that are both mathematically treatable and biologically significant, called BAN-LGT networks. Then, we study the decomposition of such networks in subnetworks with three leaves and ask whether or not this decomposition determines the network. The answer to this question is negative, but if we further impose time-consistency (species involved in a later gene transfer must coexist) the answer is affirmative, up to some redundancy that can never be recovered but is fully characterized.

Lost in space? Generalising subtree prune and regraft to spaces of phylogenetic networks

Over the last fifteen years, phylogenetic networks have become a popular tool to analyse relationships between species whose past includes reticulation events such as hybridisation or horizontal gene transfer. However, the space of phylogenetic networks is significantly larger than that of phylogenetic trees, and how to analyse and search this enlarged space remains a poorly understood problem. Inspired by the widely-used rooted subtree prune and regraft (rSPR) operation on rooted phylogenetic trees, we propose a new operation-called subnet prune and regraft (SNPR)-that induces a metric on the space of all rooted phylogenetic networks on a fixed set of leaves. We show that the spaces of several popular classes of rooted phylogenetic networks (e.g. tree child, reticulation visible, and tree based) are connected under SNPR and that connectedness remains for the subclasses of these networks with a fixed number of reticulations. Lastly, we bound the distance between two rooted phylogenetic networks under the SNPR operation, show that it is computationally hard to compute this distance exactly, and analyse how the SNPR-distance between two such networks relates to the rSPR-distance between rooted phylogenetic trees that are embedded in these networks.

A program to compute the soft Robinson-Foulds distance between phylogenetic networks

Background

Over the past two decades, phylogenetic networks have been studied to model reticulate evolutionary events. The relationships among phylogenetic networks, phylogenetic trees and clusters serve as the basis for reconstruction and comparison of phylogenetic networks. To understand these relationships, two problems are raised: the tree containment problem, which asks whether a phylogenetic tree is displayed in a phylogenetic network, and the cluster containment problem, which asks whether a cluster is represented at a node in a phylogenetic network. Both the problems are NP-complete.

Results

A fast exponential-time algorithm for the cluster containment problem on arbitrary networks is developed and implemented in C. The resulting program is further extended into a computer program for fast computation of the Soft Robinson–Foulds distance between phylogenetic networks.

Conclusions

Two computer programs are developed for facilitating reconstruction and validation of phylogenetic network models in evolutionary and comparative genomics. Our simulation tests indicated that they are fast enough for use in practice. Additionally, the distribution of the Soft Robinson–Foulds distance between phylogenetic networks is demonstrated to be unlikely normal by our simulation data.

Electronic supplementary material

The online version of this article (doi:10.1186/s12864-017-3500-5) contains supplementary material, which is available to authorized users.

A Survey of Methods for Constructing Rooted Phylogenetic Networks

Rooted phylogenetic networks are primarily used to represent conflicting evolutionary information and describe the reticulate evolutionary events in phylogeny. So far a lot of methods have been presented for constructing rooted phylogenetic networks, of which the methods based on the decomposition property of networks and by means of the incompatible graph (such as the CASS, the LNETWORK and the BIMLR) are more efficient than other available methods. The paper will discuss and compare these methods by both the practical and artificial datasets, in the aspect of the running time of the methods and the effective of constructed phylogenetic networks. The results show that the LNETWORK can construct much simper networks than the others.

Fast Construction of Near Parsimonious Hybridization Networks for Multiple Phylogenetic Trees

Hybridization networks represent plausible evolutionary histories of species that are affected by reticulate evolutionary processes. An established computational problem on hybridization networks is constructing the most parsimonious hybridization network such that each of the given phylogenetic trees (called gene trees) is "displayed" in the network. There have been several previous approaches, including an exact method and several heuristics, for this NP-hard problem. However, the exact method is only applicable to a limited range of data, and heuristic methods can be less accurate and also slow sometimes. In this paper, we develop a new algorithm for constructing near parsimonious networks for multiple binary gene trees. This method is more efficient for large numbers of gene trees than previous heuristics. This new method also produces more parsimonious results on many simulated datasets as well as a real biological dataset than a previous method. We also show that our method produces topologically more accurate networks for many datasets.

The comparison of tree-sibling time consistent phylogenetic networks is graph isomorphism-complete

Several polynomial time computable metrics on the class of semibinary tree-sibling time consistent phylogenetic networks are available in the literature; in particular, the problem of deciding if two networks of this kind are isomorphic is in P. In this paper, we show that if we remove the semibinarity condition, then the problem becomes much harder. More precisely, we prove that the isomorphism problem for generic tree-sibling time consistent phylogenetic networks is polynomially equivalent to the graph isomorphism problem. Since the latter is believed not to belong to P, the chances are that it is impossible to define a metric on the class of all tree-sibling time consistent phylogenetic networks that can be computed in polynomial time.

BIMLR: a method for constructing rooted phylogenetic networks from rooted phylogenetic trees

Rooted phylogenetic trees constructed from different datasets (e.g. from different genes) are often conflicting with one another, i.e. they cannot be integrated into a single phylogenetic tree. Phylogenetic networks have become an important tool in molecular evolution, and rooted phylogenetic networks are able to represent conflicting rooted phylogenetic trees. Hence, the development of appropriate methods to compute rooted phylogenetic networks from rooted phylogenetic trees has attracted considerable research interest of late. The CASS algorithm proposed by van Iersel et al. is able to construct much simpler networks than other available methods, but it is extremely slow, and the networks it constructs are dependent on the order of the input data. Here, we introduce an improved CASS algorithm, BIMLR. We show that BIMLR is faster than CASS and less dependent on the input data order. Moreover, BIMLR is able to construct much simpler networks than almost all other methods. BIMLR is available at http://nclab.hit.edu.cn/wangjuan/BIMLR/.

Trinets encode tree-child and level-2 phylogenetic networks

Phylogenetic networks generalize evolutionary trees, and are commonly used to represent evolutionary histories of species that undergo reticulate evolutionary processes such as hybridization, recombination and lateral gene transfer. Recently, there has been great interest in trying to develop methods to construct rooted phylogenetic networks from triplets, that is rooted trees on three species. However, although triplets determine or encode rooted phylogenetic trees, they do not in general encode rooted phylogenetic networks, which is a potential issue for any such method. Motivated by this fact, Huber and Moulton recently introduced trinets as a natural extension of rooted triplets to networks. In particular, they showed that [Formula: see text] phylogenetic networks are encoded by their trinets, and also conjectured that all "recoverable" rooted phylogenetic networks are encoded by their trinets. Here we prove that recoverable binary level-2 networks and binary tree-child networks are also encoded by their trinets. To do this we prove two decomposition theorems based on trinets which hold for all recoverable binary rooted phylogenetic networks. Our results provide some additional evidence in support of the conjecture that trinets encode all recoverable rooted phylogenetic networks, and could also lead to new approaches to construct phylogenetic networks from trinets.

Metrics on multilabeled trees: interrelationships and diameter bounds

Multilabeled trees or MUL-trees, for short, are trees whose leaves are labeled by elements of some nonempty finite set X such that more than one leaf may be labeled by the same element of X. This class of trees includes phylogenetic trees and tree shapes. MUL-trees arise naturally in, for example, biogeography and gene evolution studies and also in the area of phylogenetic network reconstruction. In this paper, we introduce novel metrics which may be used to compare MUL-trees, most of which generalize well-known metrics on phylogenetic trees and tree shapes. These metrics can be used, for example, to better understand the space of MUL-trees or to help visualize collections of MUL-trees. In addition, we describe some relationships between the MUL-tree metrics that we present and also give some novel diameter bounds for these metrics. We conclude by briefly discussing some open problems as well as pointing out how MUL-tree metrics may be used to define metrics on the space of phylogenetic networks.

A practical algorithm for reconstructing level-1 phylogenetic networks

Recently, much attention has been devoted to the construction of phylogenetic networks which generalize phylogenetic trees in order to accommodate complex evolutionary processes. Here, we present an efficient, practical algorithm for reconstructing level-1 phylogenetic networks--a type of network slightly more general than a phylogenetic tree--from triplets. Our algorithm has been made publicly available as the program LEV1ATHAN. It combines ideas from several known theoretical algorithms for phylogenetic tree and network reconstruction with two novel subroutines. Namely, an exponential-time exact and a greedy algorithm both of which are of independent theoretical interest. Most importantly, LEV1ATHAN runs in polynomial time and always constructs a level-1 network. If the data are consistent with a phylogenetic tree, then the algorithm constructs such a tree. Moreover, if the input triplet set is dense and, in addition, is fully consistent with some level-1 network, it will find such a network. The potential of LEV1ATHAN is explored by means of an extensive simulation study and a biological data set. One of our conclusions is that LEV1ATHAN is able to construct networks consistent with a high percentage of input triplets, even when these input triplets are affected by a low to moderate level of noise.

Comparison of galled trees

Galled trees, directed acyclic graphs that model evolutionary histories with isolated hybridization events, have become very popular due to both their biological significance and the existence of polynomial-time algorithms for their reconstruction. In this paper, we establish to which extent several distance measures for the comparison of evolutionary networks are metrics for galled trees, and hence, when they can be safely used to evaluate galled tree reconstruction methods.

Characterization of phylogenetic networks with NetTest

Background

Typical evolutionary events like recombination, hybridization or gene transfer make necessary the use of phylogenetic networks to properly depict the evolution of DNA and protein sequences. Although several theoretical classes have been proposed to characterize these networks, they make stringent assumptions that will likely not be met by the evolutionary process. We have recently shown that the complexity of simulated networks is a function of the population recombination rate, and that at moderate and large recombination rates the resulting networks cannot be categorized. However, we do not know whether these results extend to networks estimated from real data.

Results

We introduce a web server for the categorization of explicit phylogenetic networks, including the most relevant theoretical classes developed so far. Using this tool, we analyzed statistical parsimony phylogenetic networks estimated from ~5,000 DNA alignments, obtained from the NCBI PopSet and Polymorphix databases. The level of characterization was correlated to nucleotide diversity, and a high proportion of the networks derived from these data sets could be formally characterized.

Conclusions

We have developed a public web server, NetTest (freely available from the software section at http://darwin.uvigo.es), to formally characterize the complexity of phylogenetic networks. Using NetTest we found that most statistical parsimony networks estimated with the program TCS could be assigned to a known network class. The level of network characterization was correlated to nucleotide diversity and dependent upon the intra/interspecific levels, although no significant differences were detected among genes. More research on the properties of phylogenetic networks is clearly needed.

A metric on the space of reduced phylogenetic networks

Phylogenetic networks are leaf-labeled, rooted, acyclic, and directed graphs that are used to model reticulate evolutionary histories. Several measures for quantifying the topological dissimilarity between two phylogenetic networks have been devised, each of which was proven to be a metric on certain restricted classes of phylogenetic networks. A biologically motivated class of phylogenetic networks, namely, reduced phylogenetic networks, was recently introduced. None of the existing measures is a metric on the space of reduced phylogenetic networks. In this paper, we provide a metric on the space of reduced phylogenetic networks that is computable in time polynomial in the size of the networks.

Metrics for phylogenetic networks II: nodal and triplets metrics

The assessment of phylogenetic network reconstruction methods requires the ability to compare phylogenetic networks. This is the second in a series of papers devoted to the analysis and comparison of metrics for tree-child time consistent phylogenetic networks on the same set of taxa. In this paper, we generalize to phylogenetic networks two metrics that have already been introduced in the literature for phylogenetic trees: the nodal distance and the triplets distance. We prove that they are metrics on any class of tree-child time consistent phylogenetic networks on the same set of taxa, as well as some basic properties for them. To prove these results, we introduce a reduction/expansion procedure that can be used not only to establish properties of tree-child time consistent phylogenetic networks by induction, but also to generate all tree-child time consistent phylogenetic networks with a given number of leaves.

Characterization of reticulate networks based on the coalescent with recombination

Phylogenetic networks aim to represent the evolutionary history of taxa. Within these, reticulate networks are explicitly able to accommodate evolutionary events like recombination, hybridization, or lateral gene transfer. Although several metrics exist to compare phylogenetic networks, they make several assumptions regarding the nature of the networks that are not likely to be fulfilled by the evolutionary process. In order to characterize the potential disagreement between the algorithms and the biology, we have used the coalescent with recombination to build the type of networks produced by reticulate evolution and classified them as regular, tree sibling, tree child, or galled trees. We show that, as expected, the complexity of these reticulate networks is a function of the population recombination rate. At small recombination rates, most of the networks produced are already more complex than regular or tree sibling networks, whereas with moderate and large recombination rates, no network fit into any of the standard classes. We conclude that new metrics still need to be devised in order to properly compare two phylogenetic networks that have arisen from reticulating evolutionary process.