Accounting for spatial sampling patterns in Bayesian phylogeography

On the connections between the spatial Lambda-Fleming-Viot model and other processes for analysing geo-referenced genetic data

The introduction of the spatial Lambda-Fleming-Viot model (ΛV) in population genetics was mainly driven by the pioneering work of Alison Etheridge, in collaboration with Nick Barton and Amandine Véber about ten years ago (Barton et al., 2010; Barton et al., 2013). The ΛV model provides a sound mathematical framework for describing the evolution of a population of related individuals along a spatial continuum. It alleviates the "pain in the torus" issue with Wright and Malécot's isolation by distance model and is sampling consistent, making it a tool of choice for statistical inference. Yet, little is known about the potential connections between the ΛV and other stochastic processes generating trees and the spatial coordinates along the corresponding lineages. This work focuses on a version of the ΛV whereby lineages move rapidly over small distances. Using simulations, we show that the induced ΛV tree-generating process is well approximated by a birth-death model. Our results also indicate that Brownian motions modelling the movements of lines of descent along birth-death trees do not generally provide a good approximation of the ΛV due to habitat boundaries effects that play an increasingly important role in the long run. Accounting for habitat boundaries through reflected Brownian motions considerably increases the similarity to the ΛV model however. Finally, we describe efficient algorithms for fast simulation of the backward and forward in time versions of the ΛV model.

Modeling the velocity of evolving lineages and predicting dispersal patterns

Accurate estimation of the dispersal velocity or speed of evolving organisms is no mean feat. In fact, existing probabilistic models in phylogeography or spatial population genetics generally do not provide an adequate framework to define velocity in a relevant manner. For instance, the very concept of instantaneous speed simply does not exist under one of the most popular approaches that models the evolution of spatial coordinates as Brownian trajectories running along a phylogeny [30]. Here, we introduce a new family of models - the so-called "Phylogenetic Integrated Velocity" (PIV) models - that use Gaussian processes to explicitly model the velocity of evolving lineages instead of focusing on the fluctuation of spatial coordinates over time. We describe the properties of these models and show an increased accuracy of velocity estimates compared to previous approaches. Analyses of West Nile virus data in the U.S.A. indicate that PIV models provide sensible predictions of the dispersal of evolving pathogens at a one-year time horizon. These results demonstrate the feasibility and relevance of predictive phylogeography in monitoring epidemics in time and space.

How fast are viruses spreading in the wild?

Genomic data collected from viral outbreaks can be exploited to reconstruct the dispersal history of viral lineages in a two-dimensional space using continuous phylogeographic inference. These spatially explicit reconstructions can subsequently be used to estimate dispersal metrics allowing to unveil the dispersal dynamics and evaluate the capacity to spread among hosts. Heterogeneous sampling intensity of genomic sequences can however impact the accuracy of dispersal insights gained through phylogeographic inference. In our study, we implement a simulation framework to evaluate the robustness of three dispersal metrics - a lineage dispersal velocity, a diffusion coefficient, and an isolation-by-distance signal metric - to the sampling effort. Our results reveal that both the diffusion coefficient and isolation-by-distance signal metrics appear to be robust to the number of samples considered for the phylogeographic reconstruction. We then use these two dispersal metrics to compare the dispersal pattern and capacity of various viruses spreading in animal populations. Our comparative analysis reveals a broad range of isolation-by-distance patterns and diffusion coefficients mostly reflecting the dispersal capacity of the main infected host species but also, in some cases, the likely signature of rapid and/or long-distance dispersal events driven by human-mediated movements through animal trade. Overall, our study provides key recommendations for the lineage dispersal metrics to consider in future studies and illustrates their application to compare the spread of viruses in various settings.

Approaches and challenges to inferring the geographical source of infectious disease outbreaks using genomic data

Genomic data hold increasing potential in the elucidation of transmission dynamics and geographical sources of infectious disease outbreaks. Phylogeographic methods that use epidemiological and genomic data obtained from surveillance enable us to infer the history of spatial transmission that is naturally embedded in the topology of phylogenetic trees as a record of the dispersal of infectious agents between geographical locations. In this Review, we provide an overview of phylogeographic approaches widely used for reconstructing the geographical sources of outbreaks of interest. These approaches can be classified into ancestral trait or state reconstruction and structured population models, with structured population models including popular structured coalescent and birth-death models. We also describe the major challenges associated with sequencing technologies, surveillance strategies, data sharing, and analysis frameworks that became apparent during the generation of large-scale genomic data in recent years, extending beyond inference approaches. Finally, we highlight the role of genomic data in geographical source inference and clarify how this enhances understanding and molecular investigations of outbreak sources.

Uncovering the endemic circulation of rabies in Cambodia

In epidemiology, endemicity characterizes sustained pathogen circulation in a geographical area, which involves a circulation that is not being maintained by external introductions. Because it could potentially shape the design of public health interventions, there is an interest in fully uncovering the endemic pattern of a disease. Here, we use a phylogeographic approach to investigate the endemic signature of rabies virus (RABV) circulation in Cambodia. Cambodia is located in one of the most affected regions by rabies in the world, but RABV circulation between and within Southeast Asian countries remains understudied. Our analyses are based on a new comprehensive data set of 199 RABV genomes collected between 2014 and 2017 as well as previously published Southeast Asian RABV sequences. We show that most Cambodian sequences belong to a distinct clade that has been circulating almost exclusively in Cambodia. Our results thus point towards rabies circulation in Cambodia that does not rely on external introductions. We further characterize within-Cambodia RABV circulation by estimating lineage dispersal metrics that appear to be similar to other settings, and by performing landscape phylogeographic analyses to investigate environmental factors impacting the dispersal dynamic of viral lineages. The latter analyses do not lead to the identification of environmental variables that would be associated with the heterogeneity of viral lineage dispersal velocities, which calls for a better understanding of local dog ecology and further investigations of the potential drivers of RABV spread in the region. Overall, our study illustrates how phylogeographic investigations can be performed to assess and characterize viral endemicity in a context of relatively limited data.

Impact and mitigation of sampling bias to determine viral spread: Evaluating discrete phylogeography through CTMC modeling and structured coalescent model approximations

Bayesian phylogeographic inference is a powerful tool in molecular epidemiological studies, which enables reconstruction of the origin and subsequent geographic spread of pathogens. Such inference is, however, potentially affected by geographic sampling bias. Here, we investigated the impact of sampling bias on the spatiotemporal reconstruction of viral epidemics using Bayesian discrete phylogeographic models and explored different operational strategies to mitigate this impact. We considered the continuous-time Markov chain (CTMC) model and two structured coalescent approximations (Bayesian structured coalescent approximation [BASTA] and marginal approximation of the structured coalescent [MASCOT]). For each approach, we compared the estimated and simulated spatiotemporal histories in biased and unbiased conditions based on the simulated epidemics of rabies virus (RABV) in dogs in Morocco. While the reconstructed spatiotemporal histories were impacted by sampling bias for the three approaches, BASTA and MASCOT reconstructions were also biased when employing unbiased samples. Increasing the number of analyzed genomes led to more robust estimates at low sampling bias for the CTMC model. Alternative sampling strategies that maximize the spatiotemporal coverage greatly improved the inference at intermediate sampling bias for the CTMC model, and to a lesser extent, for BASTA and MASCOT. In contrast, allowing for time-varying population sizes in MASCOT resulted in robust inference. We further applied these approaches to two empirical datasets: a RABV dataset from the Philippines and a SARS-CoV-2 dataset describing its early spread across the world. In conclusion, sampling biases are ubiquitous in phylogeographic analyses but may be accommodated by increasing the sample size, balancing spatial and temporal composition in the samples, and informing structured coalescent models with reliable case count data.

Data integration in Bayesian phylogenetics

Researchers studying the evolution of viral pathogens and other organisms increasingly encounter and use large and complex data sets from multiple different sources. Statistical research in Bayesian phylogenetics has risen to this challenge. Researchers use phylogenetics not only to reconstruct the evolutionary history of a group of organisms, but also to understand the processes that guide its evolution and spread through space and time. To this end, it is now the norm to integrate numerous sources of data. For example, epidemiologists studying the spread of a virus through a region incorporate data including genetic sequences (e.g. DNA), time, location (both continuous and discrete) and environmental covariates (e.g. social connectivity between regions) into a coherent statistical model. Evolutionary biologists routinely do the same with genetic sequences, location, time, fossil and modern phenotypes, and ecological covariates. These complex, hierarchical models readily accommodate both discrete and continuous data and have enormous combined discrete/continuous parameter spaces including, at a minimum, phylogenetic tree topologies and branch lengths. The increased size and complexity of these statistical models have spurred advances in computational methods to make them tractable. We discuss both the modeling and computational advances below, as well as unsolved problems and areas of active research.

SPREAD 4: online visualisation of pathogen phylogeographic reconstructions

Phylogeographic analyses aim to extract information about pathogen spread from genomic data, and visualising spatio-temporal reconstructions is a key aspect of this process. Here we present SPREAD 4, a feature-rich web-based application that visualises estimates of pathogen dispersal resulting from Bayesian phylogeographic inference using BEAST on a geographic map, offering zoom-and-filter functionality and smooth animation over time. SPREAD 4 takes as input phylogenies with both discrete and continuous location annotation and offers customised visualisation as well as generation of publication-ready figures. SPREAD 4 now features account-based storage and easy sharing of visualisations by means of unique web addresses. SPREAD 4 is intuitive to use and is available online at https://spreadviz.org, with an accompanying web page containing answers to frequently asked questions at https://beast.community/spread4.

Phylodynamic analysis of the highly pathogenic avian influenza H5N8 epidemic in France, 2016-2017

In 2016-2017, France experienced a devastating epidemic of highly pathogenic avian influenza (HPAI) H5N8, with more than 400 outbreaks reported in poultry farms. We analyzed the spatiotemporal dynamics of the epidemic using a structured-coalescent-based phylodynamic approach that combined viral genomic data (n = 196; one viral genome per farm) and epidemiological data. In the process, we estimated viral migration rates between départements (French administrative regions) and the temporal dynamics of the effective viral population size (Ne) in each département. Viral migration rates quantify viral spread between départements and Ne is a population genetic measure of the epidemic size and, in turn, is indicative of the within-département transmission intensity. We extended the phylodynamic analysis with a generalized linear model to assess the impact of multiple factors-including large-scale preventive culling and live-duck movement bans-on viral migration rates and Ne. We showed that the large-scale culling of ducks that was initiated on 4 January 2017 significantly reduced the viral spread between départements. No relationship was found between the viral spread and duck movements between départements. The within-département transmission intensity was found to be weakly associated with the intensity of duck movements within départements. Together, these results indicated that the virus spread in short distances, either between adjacent départements or within départements. Results also suggested that the restrictions on duck transport within départements might not have stopped the viral spread completely. Overall, we demonstrated the usefulness of phylodynamics in characterizing the dynamics of a HPAI epidemic and assessing control measures. This method can be adapted to investigate other epidemics of fast-evolving livestock pathogens.

The Evolution, Genomic Epidemiology, and Transmission Dynamics of Tembusu Virus

Tembusu virus (TMUV) can induce severe egg drop syndrome in ducks, causing significant economic losses. In this study, the possible origin, genomic epidemiology, and transmission dynamics of TMUV were determined. The time to the most recent common ancestor of TMUV was found to be 1924, earlier than that previously reported. The effective population size of TMUV increased rapidly from 2010 to 2013 and was associated with the diversification of different TMUV clusters. TMUV was classified into three clusters (clusters 1, 2, and 3) based on the envelope (E) protein. Subcluster 2.2, within cluster 2, is the most prevalent, and the occurrence of these mutations is accompanied by changes in the virulence and infectivity of the virus. Two positive selections on codons located in the NS3 and NS5 genes (591 of NS3 and 883 of NS5) were identified, which might have caused changes in the ability of the virus to replicate. Based on phylogeographic analysis, Malaysia was the most likely country of origin for TMUV, while Shandong Province was the earliest province of origin in China. This study has important implications for understanding TMUV and provides suggestions for its prevention and control.

Epidemiological Inference From Pathogen Genomes: A Review of Phylodynamic Models and Applications

Phylodynamics requires an interdisciplinary understanding of phylogenetics, epidemiology, and statistical inference. It has also experienced more intense application than ever before amid the SARS-CoV-2 pandemic. In light of this, we present a review of phylodynamic models beginning with foundational models and assumptions. Our target audience is public health researchers, epidemiologists, and biologists seeking a working knowledge of the links between epidemiology, evolutionary models, and resulting epidemiological inference. We discuss the assumptions linking evolutionary models of pathogen population size to epidemiological models of the infected population size. We then describe statistical inference for phylodynamic models and list how output parameters can be rearranged for epidemiological interpretation. We go on to cover more sophisticated models and finish by highlighting future directions.

The impact of sampling bias on viral phylogeographic reconstruction

Genomic epidemiology plays an ever-increasing role in our understanding of and response to the spread of infectious pathogens. Phylogeography, the reconstruction of the historical location and movement of pathogens from the evolutionary relationships among sampled pathogen sequences, can inform policy decisions related to viral movement among jurisdictions. However, phylogeographic reconstruction is impacted by the fact that the sampling and virus sequencing policies differ among jurisdictions, and these differences can cause bias in phylogeographic reconstructions. Here we assess the potential impacts of geographic-based sampling bias on estimated viral locations in the past, and on whether key viral movements can be detected. We quantify the effect of bias using simulated phylogenies with known geographic histories, and determine the impact of the biased sampling and of the underlying migration rate on the accuracy of estimated past viral locations. We find that overall, the accuracy of phylogeographic reconstruction is high, particularly when the migration rate is low. However, results depend on sampling, and sampling bias can have a large impact on the numbers and nature of estimated migration events. We apply these insights to the geographic spread of Ebolavirus in the 2014-2016 West Africa epidemic. This work highlights how sampling policy can both impact geographic inference and be optimized to best ensure the accuracy of specific features of geographic spread.