Deep untargeted wastewater metagenomic sequencing from sewersheds across the United States

This paper presents CASPER, the largest publicly available untargeted wastewater metagenomic dataset to date, comprising 1,206 deep-sequenced samples from across the United States that enable population-scale pathogen surveillance, correlation with clinical trends, and the detection of emerging viral threats like avian influenza and measles.

The Gist

Imagine the entire United States has a giant, invisible nervous system made of sewage pipes. Every day, millions of people flush toilets, wash hands, and shower, sending a tiny, diluted version of their health history into this system. For years, scientists have been able to "listen" to this system, but only if they knew exactly what sound to listen for—like tuning a radio to a specific station to hear one song.

This paper introduces CASPER, a massive new project that doesn't just tune into one station. Instead, it turns the volume up to maximum and listens to every frequency at once, capturing a "deep dive" into the microbial world flowing through our cities.

Here is the story of CASPER, explained simply:

1. The Old Way vs. The New Way

The Old Way (Targeted Testing):
Imagine you are looking for a specific lost coin in a massive, muddy river. You have a metal detector, but it's only programmed to beep for quarters. If you lose a dime, a gold ring, or a rare ancient coin, your detector stays silent. This is how most wastewater testing works today. They use PCR tests to look for known viruses like SARS-CoV-2 or the Flu. It's great for those specific targets, but if a brand-new, unknown virus shows up, the detector won't know it's there.

The New Way (CASPER's Untargeted Sequencing):
CASPER is like hiring a team of super-scientists to scoop up a bucket of that river water and examine every single particle in it under a microscope. They aren't looking for a specific coin; they are cataloging every speck of dust, every grain of sand, and every piece of metal. Because they look at everything, they can spot a brand-new, unknown coin (a novel virus) the moment it appears, even if no one has ever seen it before.

2. The Scale: A Library of Life

This isn't just a small experiment. CASPER is the largest library of its kind ever built.

• The Reach: They collected samples from 27 different cities across 9 states, covering about 13 million people. That's roughly 4% of the entire US population.
• The Depth: To make sure they don't miss the "rare coins" (rare viruses), they didn't just skim the surface. They sequenced the DNA/RNA of the water so deeply that they generated 1.2 trillion pairs of genetic "letters."
• The Comparison: To put this in perspective, this single dataset makes up 66% of all the untargeted wastewater sequencing data currently available in the entire world's public database. It's like they wrote two-thirds of the world's encyclopedia on this topic in one go.

3. What Did They Find?

Because they looked at everything, they found a mix of the expected and the surprising:

• The "Background Noise": Most of what they found was bacteria and plant viruses (from the food people ate, like tomatoes and peppers). This is the "mud" in the river.
• The "Known Suspects": They successfully tracked common viruses like the Flu, RSV, and SARS-CoV-2. The data matched perfectly with hospital records and standard PCR tests, proving their "super-microscope" works just as well as the old metal detectors.
• The "Surprise Guests": This is the magic part. Because they weren't looking for specific targets, they caught viruses that were sneaking in:
  • Bird Flu (H5N1): They spotted signs of bird flu in dairy cattle wastewater in Missouri before it was widely confirmed in herds.
  • West Nile Virus: They detected this mosquito-borne virus in the water.
  • Measles: In Hawaii (data not in this specific release but mentioned as part of the network), they found measles virus in the sewage even though no sick people had been reported in the hospital yet.

4. Why Does This Matter?

Think of CASPER as an early warning system for the planet.

• No Waiting for Symptoms: People often wait until they feel sick to go to the doctor. By the time they do, they might have already spread the virus. Wastewater captures the virus the moment it leaves the body, often days before anyone feels ill.
• The "Unknown Unknowns": If a new pandemic virus jumps from animals to humans, the old metal detectors won't find it because they don't know what the "coin" looks like yet. CASPER sees it immediately because it looks at everything.
• A Shared Resource: The scientists are giving all this data away for free. It's like they built a massive, high-tech observatory and handed the keys to every scientist, doctor, and student in the world so they can study it and build better tools.

The Bottom Line

CASPER is a giant, deep-sea net cast across the US sewage system. Instead of just catching the fish we know are there, it catches everything swimming in the dark. This allows us to see the ocean of germs in our cities with unprecedented clarity, giving us a head start on stopping outbreaks before they even start. It turns the "sewage" of our cities into a crystal ball for public health.

Technical Summary

1. Problem Statement

Wastewater-based epidemiology (WBE) has emerged as a critical tool for monitoring community infections independent of clinical healthcare-seeking behavior. However, current monitoring programs rely heavily on targeted approaches (qPCR, dPCR, amplicon sequencing) which face significant limitations:

• Blindness to Novelty: They cannot detect pathogens outside their predefined targets.
• Mutation Vulnerability: Primer/probe binding sites can mutate, leading to false negatives.
• Latency: Designing and validating new assays for emerging threats is slow.
• Limited Scope: They miss unexpected or zoonotic threats.

While untargeted wastewater metagenomic sequencing (WW-MGS) offers a pathogen-agnostic solution, its adoption has been limited by:

• Low Sensitivity: Vertebrate-infecting viruses constitute a tiny fraction of total nucleic acids in wastewater. Shallow sequencing often fails to detect rare taxa.
• Cost and Scale: Deep sequencing required for sensitivity is expensive and generates massive data volumes (hundreds of terabases), requiring significant computational infrastructure.
• Data Scarcity: There is a lack of large-scale, publicly available, deep-sequencing datasets for benchmarking and research.

2. Methodology

The authors established CASPER (Coalition for Agnostic Sequencing of Pathogens from Environmental Reservoirs), a national network designed to perform deep, untargeted RNA sequencing on wastewater.

• Sampling Network:

  • Scope: 27 distinct wastewater catchments across 9 U.S. states (Northeast, Midwest, South, West), covering approximately 13 million people.
  • Sites: Includes municipal wastewater treatment plants (WWTPs), hospital sewage (grab samples), and airport manholes (Moore swabs).
  • Timeline: Samples collected from December 2023 to December 2025.
  • Volume: 1,206 samples total.

• Sample Processing & Sequencing:

  • Laboratories: Samples were processed at two primary sites: University of Missouri (MU) and SecureBio (SB). Both utilized distinct but optimized workflows for viral concentration, nucleic acid extraction, and cDNA synthesis.
  • Sequencing Depth: The study utilized Illumina NovaSeq platforms to generate ~1 billion read pairs (2x150 bp) per sample.
  • Total Output: 1.2 trillion read pairs, totaling 347 Terabases (Tb) of data. This depth is ~25-fold higher than the next deepest publicly available WW-MGS study.
  • Target: Primarily RNA viruses (including plant, bacterial, and vertebrate-infecting), though some DNA viruses are captured.

• Bioinformatics Pipeline:

  • Taxonomic Profiling: Shallow profiling (1M reads/sample) using Kraken2 with the Standard database to characterize community composition (rRNA vs. non-rRNA).
  • Deep Viral Detection: A custom Viral Metagenomics Pipeline (v3.0.1) developed by SecureBio.
    • K-mer Screening: Rapid exact k-mer matching against a masked Vertebrate-Infecting Virus (VV) database.
    • Contaminant Removal: Filtering against human, livestock, and vector genomes.
    • Alignment & Taxonomy: Bowtie2 alignment followed by a custom Lowest Common Ancestor (LCA) algorithm for taxonomic assignment.
  • Normalization: Pathogen abundance was normalized using Pepper Mild Mottle Virus (PMMoV), a stable fecal indicator, to account for variations in fecal input and sample processing.

3. Key Contributions

• Largest Public Dataset: CASPER represents the largest publicly available untargeted wastewater sequencing dataset to date, accounting for 66% of all untargeted WW-MGS data currently on the NCBI Sequence Read Archive (SRA).
• Proof of Feasibility: Demonstrates that deep, untargeted sequencing at a national scale is technically and economically feasible (cost per sample dropped from ~$4,000 in 2016 to ~$500 in 2026).
• Open Science: All raw data (BioProject PRJNA1247874) and analysis code are publicly released, serving as a resource for microbial ecology, viral discovery, and methodological development.
• Real-Time Surveillance: Integration with public dashboards and rapid reporting to public health authorities for immediate threat detection.

4. Key Results

• Taxonomic Composition:
  • rRNA Dominance: Ribosomal RNA (mostly bacterial) constitutes ~61% of reads.
  • Viral Fraction: Viruses make up a median of 1.6% of non-ribosomal reads.
  • Host Distribution: Among viral reads, plant-infecting viruses (diet-associated) dominate (58.6%), followed by unknown/ambiguous hosts (34.7%) and bacteriophages (6.0%). Vertebrate-infecting viruses (VV) represent only ~0.01% of total reads, highlighting the necessity of deep sequencing.
• Pathogen Tracking & Correlation:
  • Correlation with PCR: WW-MGS abundance trends for SARS-CoV-2, Influenza A, and RSV showed weak to strong positive correlations (Spearman R = 0.24–0.91) with CDC NWSS qPCR data.
  • Clinical Correlation: In Boston, WW-MGS data correlated strongly (R = 0.67–0.84) with clinical case counts from Massachusetts General Hospital.
  • Seasonality: The data successfully captured synchronized seasonal dynamics for respiratory (winter/spring peaks) and enteric viruses across geographically distant sites.
• Detection of Emerging Threats:
  • Avian Influenza (H5N1): Detected genotype B3.13 in Columbia, MO wastewater (March–May 2024) concurrent with USDA reports of dairy cattle infections, despite no confirmed human cases in Missouri.
  • West Nile Virus: Detected in fall 2025, informing public health messaging.
  • Measles: Identified in Hawaii (outside this specific public release) in September 2025 prior to clinical case confirmation.
  • Unexpected Signals: Detected unanticipated surges in Influenza C and various enteric viruses.

5. Significance

• Paradigm Shift: CASPER proves that the limitation of untargeted WW-MGS is not the approach itself, but the sequencing depth. By sequencing ~5,000-fold deeper than previous studies, it overcomes the "zero-read" problem for low-abundance pathogens.
• Biosecurity & Public Health: The pathogen-agnostic nature of CASPER allows for the detection of novel, unexpected, and zoonotic threats without prior knowledge of the pathogen, acting as an early warning system for pandemics.
• Scientific Resource: The dataset enables research into:
  • Viral Evolution: Tracking lineage shifts and mutations.
  • Microbial Ecology: Characterizing the "dark matter" of the urban virome.
  • Method Development: Benchmarking new AI/ML detection algorithms and hybrid-capture methods.
• Scalability: The successful deployment across diverse geographies and processing workflows demonstrates a scalable model for national-scale pathogen surveillance that can be expanded to include more sites and pathogens.

In conclusion, the paper establishes CASPER as a foundational infrastructure for next-generation public health surveillance, moving from targeted monitoring to comprehensive, deep, and untargeted pathogen discovery at a population scale.