Next-generation molecular and evolutionary epidemiology of infectious disease

Chair and organiser

Professor Andrew Rambaut University of Edinburgh UK

Spatial Phylodynamics of Emerging Infectious Disease: Rabies as a Model System

Professor Leslie Real, Emory University, USA

• Audio recording (mp3)

The application of short read next-generation-sequencing to Foot & Mouth Disease Virus

Professor Dan Haydon, University of Glasgow, UK

Abstract

Foot and mouth disease virus (FMDV) is a single positive stranded RNA virus in the family Picornaviridae that causes a serious acute acting disease of cloven-hoofed animals. The viral genome is ~8.2kb. In this presentation we will discuss analyses of a series of 20 viral isolates taken from different tissues from 4 sequentially infected cows and sequenced using Illumina technology. We will describe how the fine polymorphic structure of these virus populations changes from the inoculum used to infect the first animal, within individuals over time, and onward down the chain as a consequence of natural transmission. We will discuss two approaches to how we can distinguish ‘real’ variation from artifacts induced during sample preparation and sequencing. The first based on independently replicated sample preparation and sequencing reactions; the second based on a model of the process of sample preparation and sequencing, parameterized using sequence data in which various stages of sample preparation are omitted.

• Audio recording (mp3)

Next-generation computational statistics for infectious diseases

Professor Marc Suchard, University of California, Los Angeles, USA

Abstract

Statistical methods for comparing relative rates of synonymous and nonsynonymous substitutions maintain a central role in detecting positive selection. To identify selection, researchers often estimate the ratio of these relative rates at individual alignment sites. A reliable way to perform such estimation fits a codon-based evolutionary model that captures heterogeneity of $\dnds$ values across sites. Unfortunately, the large state space of possible codons makes codon-based models computationally prohibitive for massive data sets, containing hundreds or even thousands of sequences. Alternatives crudely estimate the numbers of synonymous and nonsynonymous substitutions at each site and use these counts to identify positively selected sites. Although these counting approaches scale well to massive data sets, they fail to account for ancestral state reconstruction uncertainty and to provide site-specific estimates. We propose a hybrid solution that borrows the computational strength of counting methods, but augments these methods with empirical Bayes modeling of synonymous and nonsynonymous substitution rates. The result is a fast and reliable method capable to identify sites under positive selection and to estimate site-specific $\dnds$ values in large data sets. Importantly, our hybrid approach, set in a Bayesian framework, integrates over the posterior distribution of phylogenies and reconstructions to quantify uncertainty about site-specific $\dnds$ estimates. Comparisons with mixture codon-based models demonstrate that this hybrid method competes well with these more principal statistical procedures and in some cases even outperforms them. We illustrate the utility of our method using human immunodeficiency virus and feline panleukopenia & canine parvovirus evolution examples.

• Audio recording (mp3)

Integrating ecological and evolutionary dynamics in spatial analysis of infectious diseases

Professor Philippe Lemey, University of Leuven, Belgium

Abstract

The influence of phylogeography is spreading throughout biology. Although different methodologies exist to study the geographical history of genetic lineages, recent developments towards model-based approaches that take a probabilistic perspective on spatiotemporal diffusion are gaining popularity in pathogen phylogeography. Such phylogenetic diffusion models, however, typically fit parameter-rich models to sparse spatial data leaving their potential for hypothesis testing largely unexploited.

Here, we demonstrate how a discrete diffusion model can be extended to simultaneously reconstruct spatiotemporal history and test the contribution of potential diffusion predictors. Using a large collection of human influenza A/H3N2 viruses sampled between 2002 and 2006, we illustrate how this approach allows for an integration of viral genetic data and human mobility measures to gain insights into influenza source sink dynamics. Finally, we examine the power of these models fitted to seasonal dynamics in predicting pandemic spread of pandemic H1N1

• Audio recording (mp3)

Chair and organiser

Dr Oliver Pybus, University of Oxford, UK

• Audio recording (mp3)

High-throughput sequencing reveals host-virus interactions during epidemics

Professor Paul Kellam, Wellcome Trust Sanger Institute, UK

Abstract

Next generation sequencing now forms a cornerstone in understanding how genetic changes in virus and host genomes influence the biological properties of viral pathogenesis, transmission and host susceptibility to infection. During 2009 and 2010, pandemic influenza swept around the world and whilst not as pathogenic as previous pandemics, influenza A H1N1/09 has caused in excess of 14000 deaths. Importantly, new infections of host populations that lack cross protective adaptive immunity allow the impact of virus and host genetic variation on virulence to be determined. Here I will describe our use of next generation sequencing and host genetic screening platforms to uncover the genetic variation of pandemic H1N1/09 over three waves of infection in the United Kingdom and determine the impact of genetic variation in human genes that control influenza infection in vivo. In particular, I will show how IFITM3 is essential for defending the host against influenza A virus in vivo and how variant IFITM3 alleles are associated with individuals hospitalised with seasonal or pandemic influenza H1N1/09 viruses. IFITM3 is under positive selection and the evolution of IFITM3 in different animal species suggests that this family of intrinsic immune effectors are essential early barriers of diverse virus infections.

• Audio recording (mp3)

Recovering transmission structure and dynamics from viral sequence data

Professor Sebastian Bonhoeffer, ETH Zurich, Switzerland

Abstract

In my talk I will cover three recent projects from my group. First, I am going to present a phylogenetic analysis of the transmission structure of the HIV epidemic in Switzerland. In particular, I will show that the phylogenetic analysis suggests that there are two subepidemics that are largely independent of each other. One subepidemic spreads among the risk group of men having sex with men, the other spreads in the heterosexual and injecting drug user risk group. The phylogenetic analysis also indicates that the impact of intravenous drug users on this epidemic has declined recently presumably as a consequence of needle exchange programs. The second part of my talk will deal with a new method that allows to infer the contact structure among HIV infected individuals from phylogenetic analysis. The method is applied to data from Switzerland and documents that there is evidence for heterogeneity in contact structure even within risk groups. Finally, I present a new phylogenetic method based on birth death processes in structured populations that allows to quantify the rates of transmission between different risk groups. The application of the developed method to data from the Latvian HIV epidemic shows among other things that the intravenous drug users transmit at a higher rate to heterosexuals then vice versa and that there is evidence for superspreaders among the risk group of men having sex with men.

• Audio recording (mp3)

Comparative spatial dynamics of acute viral infections

Dr Cecile Viboud, Fogarty International Center, NIH, USA

Abstract

A key question in disease dynamics is how infections spread in space and time. Here we will contrast models for the spatial spread of acute viral infections in the USA, with a specific focus on epidemic and pandemic influenza, respiratory syncytial virus, and rotavirus. We will assess the role of local climatic drivers, socio-demographic factors, population movements and school cycles on the spatial patterns of these viral diseases at different spatial scales.

• Audio recording (mp3)

Evolution of influenza: antigenic drift or thrift?

Professor Sunetra Gupta, University of Oxford, UK

Abstract

It is commonly believed that the degree of variability exhibited by an antigen is strongly correlated with its role in protective immunity. This dogma has recently been challenged by a theoretical study on influenza [1] which implies that epitopes of moderate variability play a crucial role. This model contrasts with the conventional “antigenic drift” hypothesis in that, because of functional constraints on the defining epitopes, the virus population is characterized by a limited set of antigenic types, all of which may be continuously generated by mutation from preexisting strains. Within this framework, influenza outbreaks arise as a consequence of host immune selection in a manner that is independent of the mode and tempo of viral mutation. I will discuss how serological data and phylogenetic studies may be used to discriminate between these two paradigms, and the implications for vaccine design.

• Audio recording (mp3)

Chair and organiser

Professor Christophe Fraser, University of Oxford, UK

Epidemiology and population genetics of influenza in a tropical setting

Dr Maciej Boni, Oxford University Clinical Research Unit in Viet Nam, Vietnam

Abstract

The epidemic dynamics and evolution of influenza A virus occur on a global scale. Influenza epidemics in temperate zones are seasonal and more predictable than transmission dynamics in tropical and sub-tropical areas, but the dynamics of these two climatic zones are closely linked, with East and Southeast (E/SE) Asia likely playing a major role in driving global influenza circulation. Influenza evolution affects the dynamics of influenza epidemics globally, but we do not know which human populations or which epidemiological conditions drive antigenic evolution in influenza. To answer this question would would need genetic and epidemiological data from regions of the world that are suspected to play a large role in global influenza dynamics. I will describe several such studies initiated in Vietnam that are aimed at understanding the circulation of human influenza viruses in Southeast Asia, as well as some of the new analytical methods we are developing to analyze the incoming data. Understanding Vietnam’s role in global influenza circulation will help us determine how important of a role E/SE Asia play in global flu dynamics, and it will help us identify which components of the Vietnam data would allow for similar analyses to be done in other Asian countries.

• Audio recording (mp3)

Phylogenomics and phylodynamics of Streptococcus pneumoniae

Professor William Hanage, Harvard School of Public Health, USA

Abstract

The pneumococcus (Streptococcus pneumoniae) is a pathogen of global significance, for which effective vaccines are available against some serotypes. Molecular epidemiologic data, both genetic and genomic, have demonstrated that pneumococci experience a relatively high rate of recombination, which shuffles loci including resistance determinants among lineages and has generated vaccine escape genotypes. Effectively identifying those regions that have undergone recombination is a crucial step in the analysis of genomic data, such that genealogies may be constructed that are not distorted by horizontal gene transfer. It is essential that methods for identifying recombination work with emerging large genomic datasets, and ad hoc methods may be preferred in this context. The results also indicate that, as previously hypothesized, some pneumococcal lineages have experienced more recombination than others.

• Audio recording (mp3)

HIV as a model phylodynamic system

Dr Simon Frost, University of Cambridge, UK

Abstract

Human immunodeficiency virus type 1 (HIV-1) is perhaps the most widely studied organism in viral phylodynamic studies, which aim to combine the epidemiology of viral transmission with the evolution of the virus, and for good reason. Not only is HIV-1 infection a significant public health issue, but a vast amount of sequence data has been generated over a period of decades, which when combined with the clock-like nature of HIV-1 evolution, allows fairly accurate reconstruction of past transmission events. Insights that have been gained by HIV-1 sequence analysis include identifying the timing of origin of HIV-1 epidemics, identifying period of exponential epidemic growth, and identification of 'transmission clusters' of infection. To date, rather simple models underlie the analysis of viral sequence data, which consider neither the detailed natural history of viral infection, nor the biased way in which sequence data is typically collected. Using HIV as an example of a model phylodynamic system, I will consider the importance of heterogeneity in geographic location, risk group, duration of infection, and age at infection in determining the structure of viral phylogenetic trees. I will also highlight how non-uniform sampling is an important confounding factor. [Joint work with Erik Volz].

• Audio recording (mp3)

The use of viral sequence data to evaluate the degree of disease superspreading, with an application to influenza H1N1pdm.

Professor Katia Koelle, Duke University, USA

• Audio recording (mp3)

Mapping the global distribution of infectious diseases: past, present and future

Professor Simon Hay, University of Oxford, UK

• Audio recording (mp3)

Chair

Professor Brian Spratt FMedSci FRS, Imperial College London, UK

The future of disease surveillance: from "feet on the street" to "clues from the cloud

Dr Larry Brilliant, Skoll Global Threats Fund, USA

Abstract

We review the history of efforts to innovate early warning of new diseases and routine reports of disease incidence and prevalence. Special attention is paid to innovations from eradication programs.. The first use of "surveillance and containment" as a program strategy led to the eradication of smallpox. The use of genomic sequencing (as in the polio eradication program) permitted assumption of sequencing of outbreaks detected by a surveillance system. "Prioxies" like chickenpox for smallpox and acute flaccid paralysis for polio allowed surveillance efforts to continue even after the target disease was eliminated. Changes in the international health regulations allowed innovations in non-governmental suveillance and case detection systems to flourish. GOARN, CORDS, Google Flu Trends, GPHIN, Health Map, ProMed and others are discussed. More recently, syndromic surveillance systems over mobile phones and self-reporting systems like Flu Near You open a new category of digital surveillance systems. From handwritten case reports, to national and WHO surveillance systems, to nongovernmental digital systems, the past has been prologue as the next century will see cloud based systems and social media like Twitter and Facebook become integrated into national and nongovernmental disease surveillance systems, with the possible result that disease detection times will plummet. Between the 1990’s and 2009, the average detection time, from outbreak start to detection, has fallen from 167 days to 20 days. Cloud based systems, social media, mobile phones, and other innovations raise the exciting possibility that it may not be long before new diseases can be detected within a short enough time to dramatically reduce the risk of pandemics, and routine disease surveillance may move from national surveillance systems to digital voluntary reporting on a large scale.

• Audio recording (mp3)

Microbial whole genome sequencing in diagnostic and public health microbiology

Professor Sharon Peacock CBE FMedSci, Professor of Public Health and Microbiology, University of Cambridge; Director, COVID-19 Genomics UK Consortium (COG-UK)

Abstract

Whole genome sequencing (WGS) provides the ultimate discrimination between closely related bacterial isolates, and the rapidly falling cost and turnaround time means that this could become a viable technology in diagnostic and reference microbiology laboratories in the near future. An obvious application for WGS is epidemiological typing to define transmission pathways of pathogens and support outbreak investigations. This talk will provide a brief outline on the potential utility of WGS for diagnostic and public health microbiology, and will then focus on the application of WGS to methicillin-resistant Staphylococcus aureus (MRSA). Data will be presented on the application of this technology to define global and local MRSA transmission, together with the results of studies in which a benchtop sequencer has been used to investigate putative MRSA outbreaks in a UK hospital.

• Audio recording (mp3)