Loading...
Organizing Committee
University of Wisconsin - Madison
University of Greifswald
Iowa State University
Delft University of Technology
Dalhousie University
Abstract
The unprecedented amount of genomic data that has become readily available presents specific challenges for the field of phylogenetic inference, which is concerned with estimating the evolutionary relationships among collections of species, populations, or sequences. These challenges include the development of evolutionary models that are sufficiently complex to be biologically realistic while remaining computationally tractable; deriving and implementing algorithms to efficiently estimate phylogenetic relationships that use models whose theoretical properties are well-understood and therefore interpretable; and devising ways to scale novel methodology developed to handle datasets that are increasingly large and complex.
This workshop focuses on statistical modeling and the scaling of phylogenetic methods. Topics will include modeling (e.g. multispecies coalescent model with extension to networks; diversification models) and inference with speed to scale to genomic datasets, consistency, and robustness using statistical, combinatorial, and algebraic approaches. This workshop will present the latest advances in these areas and serve as a forum to spark new ideas and collaborations.
This workshop focuses on statistical modeling and the scaling of phylogenetic methods. Topics will include modeling (e.g. multispecies coalescent model with extension to networks; diversification models) and inference with speed to scale to genomic datasets, consistency, and robustness using statistical, combinatorial, and algebraic approaches. This workshop will present the latest advances in these areas and serve as a forum to spark new ideas and collaborations.
Loading...
Loading participant list in background...
Workshop Schedule
Monday, October 21, 2024
8:30 - 8:50 AM EDT
Check In
11th Floor Collaborative Space
8:50 - 9:00 AM EDT
Welcome
11th Floor Lecture Hall
9:00 - 9:45 AM EDT
Open problems in comparative phylogenomics
11th Floor Lecture Hall
Speaker
Barbara Holland, University of Tasmania
Session Chair
Cécile Ané, University of Wisconsin - Madison
Abstract
A senior colleague of mine once said that "If someone published an accurate tree of life and put it on a poster just outside his door, he probably wouldn't bother to stop and look at it." David Penny didn't go quite that far, but always acknowledged that phylogenies weren't interesting in and of themselves, but rather for what they can tell us about macroevolution. These days there are a wealth of other data sources that can (and should) be integrated with phylogenies. For example, databases like TRY and AusTraits contain information on1000's of plant traits across tens of thousands of taxa. Geographical information systems allow us to match environmental data such as mean annual temperature and precipitation to species ranges. It is becoming increasingly affordable to gather transcriptomic data across multiple species. In this talk I will survey some of the ways we have been trying to use these kinds of data within the Centre for Plant Success and what we see as some of the open problems.
10:00 - 10:30 AM EDT
Coffee Break
11th Floor Collaborative Space
10:30 - 11:15 AM EDT
Parameter estimation on growing phylogenetic trees
11th Floor Lecture Hall
Speaker
Vu Dinh, University of Delaware
Session Chair
Cécile Ané, University of Wisconsin - Madison
Abstract
In recent years, there have been many efforts to understand the statistical properties of evolutionary processes on growing trees (i.e., when the number of taxa increases while the number of traits/sequence length is fixed). In most processes, correlations among the sampled data do not decay, invalidating the standard i.i.d assumptions for statistical analyses. This moves the statistical learning problem into a regime where classical tools such as the law of large numbers and the central limit theorem no longer hold, making many well-known statistical estimates inconsistent.
In this talk, I will outline the main challenges and distinctive characteristics of statistical learning of evolutionary-related correlated random variables, using the problem of estimating the transition rate of a two-state model on phylogenetic trees as the guiding example. The aim is to illustrate that in this new regime, the problems of parameter estimation and ancestral state reconstruction are inherently intertwined and may need to be solved together with the help of a new class of concentration inequalities and identifiability conditions.
11:30 AM - 12:15 PM EDT
Have we tested diversification linked to traits?- A Bayesian perspective
11th Floor Lecture Hall
Speaker
Rosana Zenil-Ferguson, University of Kentucky
Session Chair
Cécile Ané, University of Wisconsin - Madison
Abstract
Over the past 17 years, there has been a significant advancement in phylogenetic comparative models and software, aimed at exploring the relationship between diversification rates and discrete traits. In this talk, I will explain how these models represent various hypotheses, which can be more easily compared using a Bayesian framework. Additionally, I will show how employing a Bayesian framework and a graphical model approach allow us to formulate more complex cladogenesis changes based on phenotypic change.
12:30 - 2:30 PM EDT
Networking Lunch
Working Lunch - 11th Floor Collaborative Space
2:30 - 3:15 PM EDT
Inference of Phylogenetic Networks from Sequence Data using Composite Likelihood
11th Floor Lecture Hall
Speaker
Sungsik Kong, University of Wisconsin-Madison
Session Chair
Leo Van Lersel, Delft University of Technology
Abstract
While phylogenies have been essential in understanding how species evolve, they do not adequately describe some evolutionary processes. For instance, hybridization, a common phenomenon where interbreeding between two species leads to formation of a new species, must be depicted by a phylogenetic network, a structure that modifies a phylogenetic tree by allowing two branches to merge into one, resulting in reticulation. However, existing methods for estimating networks become computationally expensive as the dataset size and/or topological complexity increase. The lack of methods for scalable inference hampers phylogenetic networks from being widely used in practice, despite accumulating evidence that hybridization occurs frequently in nature. Here, we propose a novel method, PhyNEST (Phylogenetic Network Estimation using SiTe patterns), that estimates binary, level-1 phylogenetic networks with a fixed, user-specified number of reticulations directly from sequence data. By using the composite likelihood as the basis for inference, PhyNEST is able to use the full genomic data in a computationally tractable manner, eliminating the need to summarize the data as a set of gene trees prior to network estimation. To search network space, PhyNEST implements both hill climbing and simulated annealing algorithms. PhyNEST assumes that the data are composed of coalescent independent sites that evolve according to the Jukes-Cantor substitution model and that the network has a constant effective population size. Simulation studies demonstrate that PhyNEST is often more accurate than two existing composite-likelihood summary methods (SNaQ and PhyloNet) and that it is robust to at least one form of model misspecification (assuming a less complex nucleotide substitution model than the true generating model). PhyNEST is implemented in an open-source Julia package and is publicly available at https://github.com/sungsik-kong/PhyNEST.jl.
3:30 - 4:00 PM EDT
Coffee Break
11th Floor Collaborative Space
4:00 - 4:45 PM EDT
Identifiability in Phylogenetics Using Algebraic Matroids
11th Floor Lecture Hall
Speaker
Seth Sullivant, North Carolina State University
Session Chair
Leo Van Lersel, Delft University of Technology
Abstract
Identifiability is a crucial property for a statistical model since distributions in the model uniquely determine the parameters that produce them. In phylogenetics, the identifiability of the tree parameter is of particular interest since it means that phylogenetic models can be used to infer evolutionary histories from data. In this paper we introduce a new computational strategy for proving the identifiability of discrete parameters in algebraic statistical models that uses algebraic matroids naturally associated to the models. We then use this algorithm to prove that the tree parameters are generically identifiable for 2-tree CFN and K3P mixtures. We also show that the k-cycle phylogenetic network parameter is identifiable under the K2P and K3P models.
5:00 - 6:30 PM EDT
Reception
11th Floor Collaborative Space
Tuesday, October 22, 2024
9:00 - 9:45 AM EDT
TINNiK: A fast consistent method to infer the tree of blobs from gene tree quartets under the Network Multispecies Coalescent model
11th Floor Lecture Hall
Speaker
Elizabeth Allman, University of Alaska Fairbanks
Session Chair
Sungsik Kong, University of Wisconsin-Madison
Abstract
The tree of blobs of an arbitrary species network N is the tree T that displays only the tree edges in the underlying network relating taxa. Any reticulate substructures of N are collapsed to a single node in T. We show that the tree of blobs for a model network N is identifiable from quartet concordance factors under the Network Multispecies Coalescent model, and give an algorithm, TINNiK, for its consistent inference from multilocus datasets. TINNiK is implemented in the R package MSCquartets 2.0. We give examples of applying TINNiK to simulated and empirical datasets, emphasizing the role that TINNiK might play in quickly identifying locations in T where evolution may not have been tree-like. This work is joint with H. Banos, J. Mitchell, and J. Rhodes.
10:00 - 10:30 AM EDT
Coffee Break
11th Floor Collaborative Space
10:30 - 11:15 AM EDT
Cherry picking in phylogenetic trees
11th Floor Lecture Hall
Speaker
Simone Linz, University of Auckland
Session Chair
Sungsik Kong, University of Wisconsin-Madison
Abstract
Cherry-picking sequences, which are sequences on the leaves of phylogenetic trees and networks, have recently been introduced in a variety of different contexts. These include the reconstruction of phylogenetic networks from collections of phylogenetic trees and the characterisation of distances between phylogenetic trees and networks. In this talk, we discuss different types of cherry-picking sequences with a focus on investigating those that have led to recent progress in answering the following question. When can a collection of phylogenetic trees be explained by a normal network? Parts of this talk are based on joint work with Magnus Bordewich and Charles Semple.
11:30 AM - 12:15 PM EDT
Tree containment and inference for networks of bounded scanwidth
11th Floor Lecture Hall
Speaker
Mark Jones, Delft University of Technology
Session Chair
Sungsik Kong, University of Wisconsin-Madison
Abstract
Scanwidth is a recently-developed measure on directed acyclic graphs, similar in spirit to treewidth but adapted for phylogenetic networks.
In this talk I'll describe this measure and talk about some recent algorithmic results for networks of bounded scanwidth, including a fixed-parameter tractability result for the tree containment problem.
12:30 - 2:00 PM EDT
Lunch/Free Time
2:00 - 2:45 PM EDT
Cherry reduction sequences and linear extensions
11th Floor Lecture Hall
Speaker
Joan Carles Pons, University of the Balearic Islands
Session Chair
Cécile Ané, University of Wisconsin - Madison
Abstract
Orchard and tree-child networks can be completely reduced to a single node by cherry and reticulated-cherry reductions. In this talk we investigate the problem to counting these reductions. In particular, we show the connexion between computing this number and computing the number of linear extensions of the poset induced by the network. Finally, we provide an algorithm (adapted from others) to obtain this number and we emphasise that its complexity is bounded in terms of the level and the tree-width parameters of the network. This is a joint work with T. Martínez-Coronado and G. Riera.
3:00 - 3:30 PM EDT
Coffee Break
11th Floor Collaborative Space
3:30 - 5:00 PM EDT
Poster Session
11th Floor Collaborative Space
Wednesday, October 23, 2024
9:00 - 9:45 AM EDT
Using machine learning to accommodate complexity in phylogenetics
11th Floor Lecture Hall
Speaker
Megan Smith, Mississippi State University
Session Chair
Tracy Heath, Iowa State University
Abstract
Machine learning approaches have been applied across a wide range of questions in phylogenetic inference, ranging from substitution model selection to tree inference and to downstream inferences of introgression and diversification. I will highlight some of these applications and the machine learning algorithms that have been used in the field. I’ll also discuss the challenges associated with applying machine learning to questions in phylogenetics. Finally, I’ll discuss potential solutions to some of these challenges and the path forward for the successful application of machine learning to accommodate the complexity and heterogeneity that complicates phylogenetic inference.
10:00 - 10:30 AM EDT
Coffee Break
11th Floor Collaborative Space
10:30 - 11:15 AM EDT
Tropical Geometry Tools for Machine Learning and Phylogenomics
11th Floor Lecture Hall
Speaker
Ruriko Yoshida, Naval Postgraduate School
Session Chair
Tracy Heath, Iowa State University
Abstract
Linear algebra provides key tools in statistical and machine learning models. Tropical geometry can be described as the piecewise-linear analogue of classical geometry and in this talk we discuss recent exciting developments in “tropical geometric machine learning tools” in the rapidly growing area of phylogenomics, which can provide insight into relationships and evolutionary patterns of a diversity of organisms, from humans, plants and animals, to microbes and viruses. This is joint work with a subset of the following wonderful researchers: David Barnhill, George Aliatimis, John Sabol, Chris Teska, Keiji Miura, Kurt Pasque, Jefferson Huang.
11:30 AM - 12:15 PM EDT
Accommodating rate heterogeneity to substitution models
11th Floor Lecture Hall
Speaker
Marta Casanellas, Universitat Politècnica de Catalunya
Session Chair
Tracy Heath, Iowa State University
Abstract
In this talk we discuss single site substitution models used in Markov processes over phylogenetic trees. Most of the widely used models are time-continuous and implicitly consider that rates of substitution are time-homogeneous along each branch of the tree. Moreover, most of the models also assume global homogeneity by imposing a single rate matrix throughout the tree and are commonly stationary and time-reversible.
At the other end of the spectrum there is the general Markov model, which does not assume homogeneity of rates at all, nor stationarity or time-reversibility, and is the most general site substitution model one could consider on a phylogenetic tree. As it involves many parameters, it is impractical for a maximum likelihood approach and, although reconstruction methods based on phylogenetic invariants have been proposed for this model, these are not very popular yet.
In the case of amino acid substitution, models commonly used in phylogenetic software rely on empirical rates and only consider a single parameter per branch. This contrasts with the general Markov model, which involves 380 parameters per edge in this case. In between, there are the algebraic time-reversible models that we present in this talk: stationary and time-reversibile models which do not assume time-homogeneity of rates and are amenable to an algebraic approach. We will discuss how phylogenetic invariants can be adapted to these models and how they can mimic tools like the Hadamard transform. We will also discuss how restrictive continuous-time models are with respect to the general Markov.
This is joint work with R. Homs-Pons and A. Torres.
12:25 - 12:30 PM EDT
Group Photo (Immediately After Talk)
11th Floor Lecture Hall
12:30 - 2:30 PM EDT
Lunch/Free Time
2:30 - 3:15 PM EDT
From trees to networks, algorithms, techniques, and challenges
11th Floor Lecture Hall
Speaker
Norbert Zeh, Dalhousie University
Session Chair
Rosana Zenil-Ferguson, University of Kentucky
Abstract
This talk will review techniques for constructing networks from trees and their limitations. The talk will review maximum agreement forests (MAFs) and how they can be used to construct hybridization networks from 2 and 3 trees. Next, we will review kernelization and network generators for constructing hybridization networks from any number of trees. These techniques lead to significantly less efficient algorithms than the MAF-based algorithms for 2 trees but are the best known to date. The final part of the talk will consider cherry picking sequences as tools to construct temporal and tree-child networks. The talk will conclude with a number of open problems that highlight the current limitations of these techniques.
3:30 - 4:00 PM EDT
Coffee Break
11th Floor Collaborative Space
4:00 - 4:45 PM EDT
Beyond level-1: Identifying network features under a coalescent model
11th Floor Lecture Hall
Speaker
John Rhodes, University of Alaska Fairbanks
Session Chair
Rosana Zenil-Ferguson, University of Kentucky
Abstract
Hybridization, introgression, or other gene flow between species means their relationships may be better described by a network than a tree. However, inferring such a network from biological data can be complicated by a second biological process, incomplete lineage sorting, which results in gene relationships that need not match any of those displayed on the network. The Network Multispecies Coalescent Model describes the combined processes, yet both practical and theoretical problems remain in using it for inference. Most recent progress in this framework has focused only on level-1 networks (no overlapping cycles).
Considering the class of outer-labelled planar networks --- which may have any level --- we show that the circular order of taxon groups around each network blob is identifiable from such orders for 4-taxon subnetworks. Moreover, the 4-taxon order information can be identified from several types of data summaries, under models both with and without ILS. Finally, we give examples of blobs that cannot be distinguished, under many models and data types currently used for inference.
Thursday, October 24, 2024
9:00 - 9:45 AM EDT
Enumeration in mathematical phylogenetics: we are not afraid
11th Floor Lecture Hall
Speaker
Noah Rosenberg, Stanford University
Session Chair
Norbert Zeh, Dalhousie University
Abstract
The computational, mathematical, and statistical study of evolutionary trees relies on enumerative studies of discrete combinatorial structures in phylogenetics. Enumerations characterize spaces of trees and networks, assist in probabilistic computations under evolutionary models, and guide analyses of computational complexity for phylogenetic algorithms. This talk will examine the topic of combinatorial enumeration in mathematical phylogenetics, with an emphasis on new results concerning a fundamental structure: the labeled histories that arise in probability models of evolutionary trees.
10:00 - 10:30 AM EDT
Coffee Break
11th Floor Collaborative Space
10:30 - 11:15 AM EDT
Statistical analysis in Treespaces
11th Floor Lecture Hall
Speaker
Megan Owen, Lehman College, City University of New York
Session Chair
Norbert Zeh, Dalhousie University
Abstract
Sometimes we wish to analyze a set of phylogenetic trees or networks themselves, such as a set of gene trees or the output of a Bayesian inference program. One framework for doing this is to consider the trees or networks as points within a geometric moduli space that somehow encodes the non-Euclidean nature of this data. One of the most well-known of these spaces is the Billera-Holmes-Vogtmann (BHV) treespace, which contains all metric trees with the same n leaves. BHV treespace is CAT(0), or non-positively curved, allowing algorithms to quickly compute distances and Frechet means but also resulting in non-Euclidean behaviour like stickiness. In this talk, we look at current results, advantages and disadvantages, and open problems from this geometric approach to statistically analyzing sets of trees.
11:30 AM - 12:15 PM EDT
On the reliability of Maximum Parsimony for encoding and reconstructing phylogenetic trees
11th Floor Lecture Hall
Speaker
Mareike Fischer, University of Greifswald
Session Chair
Norbert Zeh, Dalhousie University
Abstract
Maximum parsimony (MP) is one of the oldest and simplest tree reconstruction criteria. While it is not based on a nucleotide substitution model but works in a purely combinatorial fashion, it is "folklore knowledge" that it works well whenever the number of substitutions is relatively small. Proving this assertion, which in some regard can be viewed as an extension of the famous Buneman theorem, is mathematically quite intriguing. In my talk, I will provide some first steps in this regard, and I will make use of some beautiful combinatorial properties of MP. The results presented in my talk can be regarded as an important step towards proving that MP is justified whenever the number of substitutions is sufficiently small. I will also highlight how these findings on MP impact Maximum Likelihood, another famous tree reconstruction criterion. I will conclude my talk by pointing out some areas for future research.
12:30 - 2:30 PM EDT
Open Problems Lunch
Working Lunch - 11th Floor Collaborative Space
2:30 - 3:15 PM EDT
Phylogenetic methods for quantitative trait mapping with complex data sets
11th Floor Lecture Hall
Speaker
Katie Thompson, University of Kentucky
Session Chair
Megan Owen, Lehman College, City University of New York
Abstract
Recent developments in association mapping methods along with improvements in sequencing technology have made it possible to link locations along the genome (single nucleotide polymorphisms, or SNPs) with quantitative traits in genome-wide association study (GWAS) data. Although this goal is central in the biological sciences, progress has been limited by the inability of existing methods to consider complex, but relevant, scenarios such as the simultaneous influence of genetic and external effects on quantitative trait(s) under study. Traditional statistical modeling (e.g., regression-based methods) are computationally feasible and perform well in detecting external effects but may miss weaker genetic signals since they fail to consider uneven evolutionary relatedness among samples. For this reason, it is difficult to detect and localize SNPs associated with quantitative traits in GWAS using classical statistics. This talk will include a coalescent approach to search for SNPs associated with quantitative traits in GWAS data by considering the evolutionary history among SNPs using phylogenetic analysis, through both mixed model and Bayesian approaches. Performance of phylogenetic and classical methods will be evaluated using simulation data. An application to a deer mouse data set example to identify SNPs associated with tail stripe color variation will also be presented to demonstrate utility of phylogenetic analysis methods in practice.
3:30 - 4:00 PM EDT
Coffee Break
11th Floor Collaborative Space
4:00 - 4:45 PM EDT
PickMe: Sample Selection for species tree reconstruction
11th Floor Lecture Hall
Speaker
Joe Rusinko, Hobart and William Smith Colleges
Session Chair
Megan Owen, Lehman College, City University of New York
Abstract
After collecting large data sets for phylogenomics studies, researchers must decide which genes or samples to include when reconstructing a species tree. Incomplete or unreliable data sets make the empiricist's decision more difficult. Researchers rely on ad hoc strategies to maximize sampling while ensuring sufficient data for accurate inferences. An algorithm called PickMe formalizes the sample selection process, assuming that the samples evolved under the Tree Multispecies Coalescent model. PickMe is implemented in the Julia programming language available at https://github.com/jrusinko/PhyloPickMe.
Friday, October 25, 2024
9:00 - 9:45 AM EDT
New bijections between phylogenetic and set-theoretic objects.
11th Floor Lecture Hall
Speaker
Andrew Francis, University of New South Wales
Session Chair
Mareike Fischer, University of Greifswald
Abstract
I will walk through some recently developed extensions to some bijections between phylogenetic trees and partitions of finite sets, that lift to forests and phylogenetic networks. These give new ways to characterise properties of networks, and potentially provide new tools to address open questions.
10:00 - 10:30 AM EDT
Coffee Break
11th Floor Collaborative Space
10:30 - 11:15 AM EDT
Beyond the Coalescent: Probabilistic Models of Gene Trees
11th Floor Lecture Hall
Speaker
Sebastien Roch, University of Wisconsin-Madison
Session Chair
Mareike Fischer, University of Greifswald
Abstract
The estimation of species trees or networks is complicated by the fact that genes evolve under biological phenomena that produce discordant histories. They include incomplete lineage sorting (ILS), gene duplication and loss (GDL), and horizontal gene transfer (HGT), all of which can be modeled using random gene tree distributions based on well-known stochastic processes (the coalescent, branching processes, random rearrangements). While ILS has been extensively studied, fewer theoretical results have been derived for GDL and HGT. In this talk, I will describe what is known about the analysis of the corresponding probabilistic models in the context of species tree estimation, including some recent results on identifiability and sample complexity.
11:30 AM - 12:15 PM EDT
Maximum Common Contractions Between Phylogenetic Networks
11th Floor Lecture Hall
Speaker
Manuel Lafond, Université de Sherbrooke
Session Chair
Mareike Fischer, University of Greifswald
Abstract
I will present recent results on the comparison of phylogenetic networks based on edge contractions and expansions, as originally proposed by Robinson and Foulds for trees. These operations connect the space of all phylogenetic networks on the same set of leaves, even if we forbid contractions that create cycles. The minimum number of contractions and expansions required to transform one network into another can therefore be used as a similarity measure. Moreover, the maximum common contraction between two networks can extract structural elements shared by two networks, which can provide valuable biological insight.
In this talk, I will discuss differences between the operational distance and common contractions - the former defines a metric space but not the latter. We also provide algorithmic results on finding common contractions, namely NP-hardness even in restricted networks, but positive results on weakly-galled networks, a generalization of galled trees.
12:30 - 2:30 PM EDT
Lunch/Free Time
3:00 - 3:30 PM EDT
Coffee Break
11th Floor Collaborative Space
