From discovery to dynamics: linking metaviromics with epidemiological inference
Metaviromic sequencing generates rich data on viral communities in wildlife, but these data are rarely used beyond virus discovery. In this talk, I will discuss how metaviromic data could be leveraged to infer epidemic dynamics in wildlife reservoir populations such as rodents.
As part of this work, we have been pursuing two methodological strands. The first focuses on quantifying viral loads from metaviromic data using hybrid-capture sequencing. The second introduces a statistical framework that estimates epidemic trajectories using longitudinally sampled metaviromic data. Through simulations, we are testing how factors such as sampling design, infection dynamics, and transmission patterns shape our ability to reliably recover epidemic signals.
By bridging metaviromic surveillance with epidemiological inference, this framework has the potential to improve our ability to monitor and predict viral dynamics in understudied wildlife reservoirs.
A model virome for a model host family
Metagenomic surveys have revolutionized our knowledge of virus diversity and evolution, and more than 150 different viruses have now been reported just from the Drosophilidae. However, historically, many metagenomic studies—including those of Drosophila—simply sequence large pools of wild flies. As such, it is hard to draw biological inferences: How does prevalence vary over time and space? How common are co-infections? Are viruses specialists or generalists? Are these viruses pathogenic, commensal, or beneficial? Are these even really infections of flies? I will summarize what is known about the virus community of Drosophilidae from large-scale pooled metagenomic surveys, focusing on more recent work on DNA viruses. Then I will illustrate how we have used individual-level PCR surveys and single-fly meta-transcriptomic studies to learn more about virus genomes and the epidemiology of virus infection. Finally, I will discuss the likely impact of viral infections on their hosts.
Challenges and opportunities in wastewater viromics
Wastewater contains a remarkable record of human, animal and environmental viral diversity. Advances in sequencing technologies and computational methods have transformed our ability to characterise viromes, creating opportunities for pathogen discovery and surveillance. Yet as our capacity to detect viruses continues to grow, so too does a fundamental challenge: how do we distinguish meaningful signals from the vast background diversity and data complexity present in wastewater?
I will begin by exploring this challenge through the lens of viral discovery. Drawing on semi-agnostic viral detection approaches, I will discuss how diverse, and sometimes unexpected, viral signals emerge from complex environmental settings. These approaches broaden our understanding of viral ecology and can reveal previously unrecognised pathogens, but they also raise questions about interpretation. Detecting a virus is only the first step; determining its source and public health relevance is considerably more challenging.
Building on these lessons, I will then focus on how wastewater surveillance can be used for public-health decision-making. Then, using examples from our evaluation of wastewater monitoring for zoonotic influenza, I will also highlight the importance of rigorous assay validation and integration with public health data.
The promise of wastewater viromics is immense. Ultimately, the value of wastewater viromics will depend not on how many viruses we can detect, but on how effectively we can identify the signals that matter and convert them into meaningful information.
Beyond discovery: using metaviromic data for ecological hypothesis testing and public health planning
Advances in next generation sequencing and data mining are producing an unprecedented volume of viral sequence data derived from animal surveillance, wastewater and clinical settings. Newly discovered viruses now arrive in bulk, revealing novel host-virus associations and reshaping our understanding of viral evolution. Yet these data also offer largely untapped opportunities to investigate the contemporary ecology and epidemiology of viruses, knowledge that is vital for infectious disease prevention and control. Using case studies from bats in Peru and marine mammals in Scotland, I will first illustrate how metaviromic data can test ecological hypotheses; for example, how host demography and environmental context shape viral richness and community composition. Next, I will show how pairing metagenomic data with machine learning can predict whether a virus is likely to infect humans, offering a way to triage high-risk viruses for research, surveillance or therapeutic development at the time of discovery. Finally, I will speculate on how these approaches could scale to match the growth of sequencing. In the future, repurposing vast global sequence repositories could allow us to predict how zoonotic risk varies across host groups, geography and environmental context, providing a path for metaviromic data to inform spillover prevention.
Recovery of microbial and viral ecophysiology linked to carbon accrual functions in peatlands under restoration
Peatlands are water-logged ecosystems that limit microbial decomposition, making them effective carbon sinks. Drainage removes these constraints on decomposition, switching them to carbon sources. Restoration aims to reverse these trends. Microorganisms influence carbon fluxes but how they respond to peatland restoration is poorly understood. The role of viruses—key regulators of microbial communities—in peatlands remains unknown.
We used metagenomics to study microbial and viral communities and quantified microbial growth rates using isotope labelling in peat soils across seven sites in Britain, each with restored, damaged, and natural peatlands. A peatland ecosystem health index, estimated from measurements of peat moisture, oxygen, pH, organic matter chemistry, and moss cover, explained a significant amount of variation in microbial and viral taxonomy and metabolism.
Viruses were geographically widespread yet exhibited ecosystem health-specific endemism. There were hints of stable “piggyback-the-winner” virus-host relationships. Microbial growth rate reduced as ecosystem health improved, underpinned by a shift towards energetically less favourable metabolic pathways. This metabolic shift in healthier peatlands can be linked to lower organic matter decomposition and carbon accrual. We demonstrate that microbial and viral ecology predictably changes with peatland health providing valuable insights for restoration targets and monitoring to maximize carbon accrual in peatlands.
Follow Ashish on Bluesky and LinkedIn
From arbovirus preparedness to clinical surveillance: metaviromics in UK veterinary contexts
Although metagenomic approaches are becoming established in human clinical and public health surveillance, their application in veterinary systems remains comparatively underdeveloped. Yet characterising viral diversity across animal populations is fundamental to improving disease surveillance, preparedness, and our ability to recognise virus emergence.
I will present investigations spanning two contrasting veterinary systems in the United Kingdom. The first uses remnant material from a recent national-scale survey of mosquitoes, which revealed multiple novel viruses and several detections of arbovirus-related lineages associated with disease in birds. The second examines diarrhoeic dogs submitted for routine diagnostic testing, showing that clinically relevant enteric viral diversity may be under-detected by standard pathogen-specific assays, and that genome-scale surveillance can help refine diagnostic targets and prioritise viruses for epidemiological follow-up.
Together, these studies show how metaviromics can detect both established and emerging viral threats in a changing UK landscape: from arboviruses now appearing in regions where they were historically absent, to viruses detected in patients that may be missed when diagnostic testing remains fixed to historical expectations.
Preferential sharing of virus populations between mothers and infants in early life
In this work, we aimed to investigate the contribution of human maternal gut virome to early-life virome assembly and adaptation in infants, and to determine patterns of viral sharing, persistence, and selection in healthy longitudinal samples of 15 mother–infant pairs from the Pregnancy and Early Life - PEARL cohort.
Viral-like particles were enriched from 137 faecal samples spanning mothers’ first trimester to infants at 2 years. Viral nucleic acids were sequenced, contigs assembled into vOTUs, taxonomically classified and maternal influence was assessed.
We identified 3531 extracellular vOTUs across the cohort. Maternal virome shared with infants showed stronger signatures of positive selection. Viral sharing and persistence were associated with specific viral families and their predicted hosts, for instance, Microviridae were enriched among persistent vOTUs, while Anelloviridae were enriched among transient ones. Overall, the infant virome is dominated by abundant, unique, transient viruses, with a smaller but consistent set of persistent shared fractions. Infant virome undergoes a major compositional shift at Month 4. Persistence of the extracellular virome in the cohort appears independent of viral lifestyle. Specific viral families and their hosts are associated with persistence and transience, with distinct patterns of sharing and non-sharing. These findings indicate that maternal–infant virome sharing is not random but structured by both host and viral lineage.
Follow Revathy on LinkedIn
Viral metagenomics for pathogen detection & characterisation
The microbial diversity in wildlife is highly understudied globally and includes pathogens that may threaten animals and human health. Next-generation sequencing (NGS) technology provides a robust platform to detect and analyse genetic material of pathogens, enabling the identification of existing or emerging diseases in an unbiased manner while complementing targeted molecular approaches.
This project investigates the viromes and broader microbiomes of invasive wildlife species, potential pathogen reservoirs, and vectors in the United Kingdom using metagenomic sequencing to better understand transmission and spillover dynamics. Within the Genomics for Animal and Plant Disease Consortium (GAP-DC), the project contributes to the development of pathogen-agnostic and targeted genomic approaches for detecting emerging biological threats to animal and plant health.
The resulting datasets will establish baseline microbial diversity, identify known and novel pathogens, and improve understanding of pathogen transmission pathways between wildlife, livestock, and humans.
Follow Mirjam on LinkedIn
Spatiotemporal characterization of the fecal virome in wild rodents from the United Kingdom
Rodents are important viral reservoirs worldwide. Understanding viral communities is complex, as multiple factors influence the prevalence and abundance of pathogenic and non-pathogenic taxa. However, studies that have attempted to explain this factor in rodent virome studies are still scarce. Our aim was to determine how viral communities are shared across sites, rodent species, and seasons, and how diversity, prevalence, and abundance are shaped by these factors. We sampled rodents at four sites in the UK, focusing on two species, Apodemus sylvaticus and Mus musculus, with two sites sampled at multiple time points. Using a metagenomic approach on rodent faeces, we characterized viral communities and applied a viral operational taxonomic unit (vOTU)-based framework to evaluate diversity patterns.
The abundance and prevalence of vertebrate-associated viral genera varied among sites, host species, and sampling time points. Diversity analyses suggested clustering predominantly by rodent species, with site-related clustering observed only forA. sylvaticus, whereas M. musculus samples showed no clear spatial differentiation. We also detected potential seasonal clustering based on vOTU abundance and presence. Through time, vertebrate-associated and total viral genera displayed substantial variation, including marked turnover during the winter–autumn transition, with notable gains and losses of taxa. Our results suggest a possible effect of seasonality and a strong influence of host species on viral community composition and diversity.
Follow Luisa on LinkedIn
Escaping the matrix: from soil elution to host-viral ecology
Soil environments harbour immense, yet largely uncharacterised, viral diversity. Our lab focuses on the component of this virome that infects bacteria: the bacteriophages. Field-scale agricultural trials offer the chance to explore the ecology of these viruses in natural microbial communities, and better understand how they regulate bacterial populations. I’ll present ongoing work from our lab where we use viromics to describe the viral community ecology, with relevance to regenerative agricultural practices. I’ll also present our attempts to link viruses to their unculturable hosts using Hi-C metagenomics, discussing challenges and opportunities of this powerful technique that physically pairs viral and host chromosomes in situ. Combined with in-lab soil manipulation experiments, we aim to identify the ultimate drivers of viral life-history evolution.
Follow Sean on Bluesky
Rodent-associated virus communities in restored woodlands influenced by host species, woodland features and landscape context
Woodland restoration is crucial for biodiversity recovery and climate mitigation, yet its influence on wildlife pathogens, and the resulting impacts on public health, remains poorly understood. While habitat succession and connectivity are known to shape host ecology, evidence of how these processes shape host-pathogen systems is limited.
To address this gap, we used metagenomics to characterise viral communities from 32 pooled samples of bank voles (Clethrionomys glareolus) and wood mice (Apodemus sylvaticus) across 16 restored Scottish woodlands varying in age (30->250 years) and connectivity. We identified 142 distinct viral operational taxonomic units (vOTUs) across 26 families of vertebrate-associated viruses, and examined how host, landscape, and woodland characteristics relate to patterns of viral abundance, richness, and community composition.
Our findings suggest that woodland restoration can reshape rodent virus communities, with woodland age, habitat structure, and landscape context all playing a role in shaping viral abundance patterns, with potential implications for transmission dynamics and zoonotic risk in restored landscapes. In contrast, variation in virus richness and community composition remained largely unexplained, pointing to the influence of local ecological context and stochastic processes. These results advance our understanding of the critical links between habitat recovery, host-virus ecology, and infectious disease risk in restored ecosystems.
From detection to inference: statistical challenges in viral community ecology
Molecular tools and sequencing have greatly expanded our ability to detect viruses in natural populations. However, the biological phenomena we ultimately want to understand, such as viral genetic diversity or interactions between viruses, are rarely observed directly. When they are not observed directly, estimating numeric descriptions of these biological phenomena from data requires statistical assumptions. The plausibility of these assumptions depends on the specifics of the system and sample collection.
I will illustrate some problems generated by incomplete observation using work on viruses in wild bumblebee communities, and show how they can or cannot be reasonably addressed statistically. Particular examples include: relating sequencing and prevalence datasets from similar but not completely overlapping samples to explore viral population genetics; and the complexities of disentangling biological interactions between viruses from shared determinants of exposure and infection risk.
The examples above illustrate a general point: when the sampling design is not suited to understanding the biological phenomena of interest, statistical analysis necessarily depends more heavily on modelling assumptions. Ideally, studies should begin by defining the research question and designing a suitable sampling strategy accordingly. However, in practice, we are limited by what has been historically collected or can feasibly be collected under resource constraints, and therefore need to consider carefully what assumptions can reasonably be imposed to answer the questions we care about. This work shows some approaches to identifying reasonable assumptions and how to apply them to a real-world example.
When virus discovery is not enough: insights and challenges beyond sequence discovery
Our view of the virus biodiversity is shaped by data availability. While eukaryotic biodiversity sampling remains highly uneven across hosts and environments, it also affects virus known diversity. Large-scale virus discovery projects have recently dramatically expanded known RNA virome. While discovery is a breakthrough moment, it is just the first step towards understanding
virus molecular biology, host range, taxonomy and interpretation.
Currently we need to improve sequence annotation and data curation. I will use my work on Polycipiviridae to illustrate how a virus family can expand into a highly divergent group with unusual genome organisations and host associations as well as unknown protein domains. I will share the examples of sequences from “viral dark matter”: novel molecular biology, taxonomic
clades and hosts. These examples will show how much can be uncovered by annotation and inform future research.
Characterization of viral dark matter needs broader ecological sampling, sequencing, and manual curation effort. At the same time, our field could take some inspiration from bacterial metagenomics. I will share the best practices of how these challenges are systematically approached at MGnify, EBI - where large scale data and analyses are being used to improve protein annotation.
Follow Ingrida on Bluesky and LinkedIn
Do host traits predict rodent virome diversity?
Host traits such as body size, lifespan, and reproductive investment have been hypothesised to affect virome diversity in wildlife, implying that some hosts may be innately more likely to host potentially pathogenic viruses than others. This project uses publicly available non-specific RNA sequencing datasets from rodents to generate a minimally-biased assessment of virus diversity and tests associations of virus richness, community composition, and individual virus family detection probability with host traits.
Title and abstract TBA