Comprehensive Wastewater Sequencing Reveals Community and Variant Dynamics of the Collective Human Virome

Michael J. Tisza, Sara Javornik Cregeen, Vasanthi Avadhanula, Ping Zhang, Tulin Ayvaz, Karen Feliz, Kristi L. Hoffman, Justin R. Clark, Austen Terwilliger, Matthew C. Ross, Juwan Cormier, David Henke, Catherine Troisi, Fuqing Wu, Janelle Ríos, Jennifer Deegan, Blake D. Hansen, John Balliew, Anna Gitter, Kehe Zhang, Runze Li, Cici Bauer, Kristina D. Mena, Pedro A. Piedra, Joseph F. Petrosino, Eric Boerwinkle, Anthony W. MaressomedRxiv 2023 · DOI

TL;DR

Researchers have moved beyond tracking single pathogens like SARS-CoV-2 in sewage to sequencing the entire "human virome." By using a probe-based enrichment technique, they achieved a 3,374-fold increase in viral signal, allowing them to track over 450 distinct viruses and monitor how viral variants evolve and spread across different cities in real-time.

What Problem Does This Paper Address?

Wastewater-based epidemiology (WBE) gained massive public attention during the COVID-19 pandemic. However, most current methods rely on PCR, which is "blind" to anything it isn't specifically designed to find. While we can track SARS-CoV-2 or Polio with high precision, we often miss the broader "virome"—the collective community of viruses circulating in a population.

Standard metagenomic sequencing (sequencing everything in a sample) usually fails in wastewater because the samples are overwhelmed by bacterial and human DNA; viruses often make up less than 0.01% of the total genetic material. This paper addresses the need for a scalable, comprehensive way to "fish out" viral sequences from the noise to understand community-wide infection dynamics and viral evolution.

Methods

The researchers developed a high-throughput workflow centered on hybrid-capture enrichment.

  • The "Bait": They used the TWIST Comprehensive Viral Research Panel, which contains over one million unique probes (molecular "hooks") designed to bind to the genomes of 3,153 different human and animal viruses.
  • The Pipeline: They created a custom bioinformatics tool called EsViritu. This pipeline handles the complex task of mapping short sequencing reads to a massive database of viral genomes, filtering out false positives, and de-replicating similar sequences to provide an accurate count of which virus strains are present.
  • Validation: To ensure the sequencing data reflected reality, they compared their results against traditional qPCR measurements and clinical "case count" data from public health departments.

Datasets

The study analyzed 363 longitudinal wastewater samples collected over nearly a year (May 2022 to February 2023).

  • Geographic Scope: Samples came from ten distinct wastewater treatment plants across two major, climatically different Texas cities: Houston (humid, coastal, international port) and El Paso (dry, inland, major border crossing).
  • Reference Data: The team built a custom Virus Pathogen Database (v2.0.2) using all available complete isolate genomes from NCBI GenBank for the genera covered by their probe set.
  • Clinical Benchmarks: They utilized publicly available clinical datasets for SARS-CoV-2, Influenza A, and Monkeypox (Mpox) to correlate wastewater signals with actual hospital and clinic reports.

Key Insights and Ideas

  • The Power of Enrichment: Without probes, the researchers found almost zero mammalian viruses. With the hybrid-capture probes, they saw a 3,374-fold enrichment, detecting between 42 and 128 different mammalian viruses in every single sample.
  • Geography is Destiny: Viral communities are not random. The "virome" of Houston looked distinctly different from that of El Paso. For example, Houston had a higher prevalence of Caliciviridae (like Norovirus), while El Paso showed stronger signals for Parvoviridae.
  • Wastewater as a Time Machine: By looking at the "Bray-Curtis dissimilarity" (a measure of how different two communities are), they found that viral populations change predictably over time. Samples taken a week apart are very similar, while those taken months apart are vastly different, reflecting the seasonal "ebb and flow" of human illness.
  • Variant Tracking: For highly abundant viruses like Astrovirus MLB1, the researchers could actually track specific mutations. They found that Astrovirus variants were geographically partitioned—Houston had its own "flavor" of the virus that evolved differently than the one in El Paso. In contrast, the JC Polyomavirus (which causes persistent, often asymptomatic infections) showed no such geographic structure, suggesting it circulates more stably and universally.

Key Results

  • Breadth of Detection: The study identified 465 distinct virus strains across the 10 sites.
  • Clinical Correlation: The sequencing data (measured in RPKMF—Reads Per Kilobase of transcript per Million filtered reads) showed strong positive correlations with clinical data. For example, the correlation ($R$) for Influenza A was 0.9, and for SARS-CoV-2, it ranged from 0.59 to 0.78.
  • Mpox Detection: The system successfully detected the 2022 Mpox outbreak in Houston wastewater, even though clinical cases were relatively low (~1,050 cases in a massive metro area).
  • Diversity Saturation: They found that viral diversity increases with the size of the population served by a treatment plant but levels off once the "catchment" area reaches about 100,000 people.

Limitations and Future Work

The primary limitation is sequencing depth. While the probes are excellent at finding viruses, some pathogens (like SARS-CoV-2) are shed in such low quantities in stool and urine that the researchers couldn't always get enough "coverage" to perform detailed variant analysis. They were able to track variants for Astrovirus and Adenovirus, but not consistently for every pathogen.

Future work will likely focus on increasing sequencing depth and refining the "bait" sets to better capture emerging threats. The authors also suggest that this method could be used to study "cryptic" transmission—viruses circulating in the community that never result in a doctor's visit or a formal clinical report.

Summary updated

Abstract

Abstract Wastewater is a discarded human by-product but analyzing it may help us understand the health of communities. Epidemiologists first analyzed wastewater to track outbreaks of poliovirus decades ago, but so-called wastewater-based epidemiology was reinvigorated to monitor SARS-CoV-2 levels. Current approaches overlook the activity of most human viruses and preclude a deeper understanding of human virome community dynamics. We conducted a comprehensive sequencing-based analysis of 363 longitudinal wastewater samples from ten distinct sites in two major cities. Over 450 distinct pathogenic viruses were detected. Sequencing reads of established pathogens and emerging viruses correlated to clinical data sets. Viral communities were tightly organized by space and time. Finally, the most abundant human viruses yielded sequence variant information consistent with regional spread and evolution. We reveal the viral landscape of human wastewater and its potential to improve our understanding of outbreaks, transmission, and its effects on overall population health. One-Sentence Summary Wastewater contains a trickle of circulating human viruses, and now we can sequence them to better track community transmission.
BibTeX
@misc{tisza2023comprehensive,
  author = {Michael J. Tisza and Sara Javornik Cregeen and Vasanthi Avadhanula and Ping Zhang and Tulin Ayvaz and Karen Feliz and Kristi L. Hoffman and Justin R. Clark and Austen Terwilliger and Matthew C. Ross and Juwan Cormier and David Henke and Catherine Troisi and Fuqing Wu and Janelle Ríos and Jennifer Deegan and Blake D. Hansen and John Balliew and Anna Gitter and Kehe Zhang and Runze Li and Cici Bauer and Kristina D. Mena and Pedro A. Piedra and Joseph F. Petrosino and Eric Boerwinkle and Anthony W. Maresso},
  title = {{Comprehensive Wastewater Sequencing Reveals Community and Variant Dynamics of the Collective Human Virome}},
  year = {2023},
  doi = {10.1101/2023.05.03.23289441},
  howpublished = {medRxiv},
}
Figure