Included in the June 5, 2025, biweekly update
This week's articles by MSU faculty, specialists and students making a difference feature virus tracking in wastewater, COVID-19 testing, AI forecasting, and U.S. ocean sampling.
In-depth comparison of untargeted and targeted sequencing for detecting virus diversity in wastewater
Li et al. (2025) compared different ways of detecting viruses in wastewater to improve how communities monitor public health threats. Wastewater carries viruses shed by people, making it a valuable tool for spotting outbreaks before symptoms are present or people get tested. But, because viruses can be hard to detect in complex samples, researchers need to understand which methods work best.
This study tested three “untargeted” sequencing methods, which scan broadly for any viruses present, and one “targeted” method that searches specifically for known human viruses. The researchers also evaluated two extraction kits and four software tools used to identify viruses from genetic data.
The findings showed strengths and limitations for each method. The targeted method was better at detecting certain gut-related viruses like norovirus and rotavirus, while the untargeted methods picked up a wider variety of human viruses overall, including some that the targeted method missed. Altogether, the researchers found 45 human-related viruses, with some only appearing through one approach. To double-check their results, they used a separate test called ddPCR, which confirmed the presence of key viruses like SARS-CoV-2 and RSV.
The researchers concluded that no single method can capture the full range of viruses in wastewater. A combined approach—using both targeted and untargeted sequencing, along with confirmatory testing—offers the most complete and reliable way to track viruses in wastewater and catch emerging threats early.
Li, Y., Bhatt, P., & Xagoraraki, I. (2025). “In-depth comparison of untargeted and targeted sequencing for detecting virus diversity in wastewater.” Water Research, 283, 123803. https://doi.org/10.1016/j.watres.2025.123803
Corresponding author: Irene Xagoraraki, xagorara@msu.edu.
Culturally Targeted Messaging and Racial Equity in SARS-CoV-2 Antibody Testing by Multiplex Salivary Measurement: Protocol Overview of a SeroNet Investigation
Lucas et al. (2025) report findings from a study aimed at improving racial equity in COVID-19 antibody testing through culturally targeted messaging and saliva-based testing. Conducted in Flint, Michigan as part of the National Institutes of Health’s SeroNet initiative, the study focused on addressing long-standing barriers that have limited the participation of African Americans in biomedical research, particularly mistrust of the medical system and discomfort with invasive testing methods.
Researchers recruited over 1,000 participants and tested whether culturally tailored outreach (communication designed for a specific community), informed by community input and offered in multiple languages, would lead to higher participation in antibody testing compared to generic public health messaging. All participants received non-invasive saliva antibody tests, and many also completed surveys about their experiences with COVID-19, health systems, and medical research.
The researchers found that culturally targeted messaging led to higher enrollment and testing completion rates among African American participants. This suggests that adapting outreach to reflect specific cultural contexts can meaningfully increase participation and trust in public health research.
The study offers a model for inclusive research during public health emergencies and beyond.
Lucas, T., Heaney, C. D., Granger, S. W., Key, K. D., Lapinski, M. K., Jones, N., Paneth, N., Brincks, A. M., Dawadi, A., Maschino, L., Rose, L., Summers, M., Weisbrod, R., Pisanic, N., Aspiras, O., Goetz, S. M. M., & Granger, D. A. (2025). “Culturally targeted messaging and racial equity in SARS-CoV-2 antibody testing by multiplex salivary measurement: Protocol overview of a SeroNet investigation.” The Lancet Regional Health – Americas, 32, 100790. https://doi.org/10.1016/j.bbih.2025.101019
Corresponding author: Todd Lucas, lucastod@msu.edu.
The FUTURE of the US Marine Seafloor and Subseafloor Sampling Capabilities
Appelgate et al. (2025) present a roadmap for strengthening the United States’ ability to collect samples from the ocean floor and the layers beneath it, which is essential for tackling major environmental, climate, and geohazard challenges. Despite the U.S. being a global leader in marine science, the researchers argue that the nation is falling behind when it comes to physical sampling capabilities. Much of the current infrastructure is aging, fragmented, or overly reliant on international programs.
The researchers—representing a national working group of marine scientists and technical experts from nearly 60 institutions—outline the urgent need to modernize U.S. seafloor and subseafloor sampling systems. These include drills, coring platforms, and mobile systems that can collect deep sediment and rock samples. Without these tools, critical research into earthquakes, carbon storage, past climate patterns, and marine ecosystem change is limited.
Appelgate et al. recommend coordinated investment from federal agencies, better integration across academic institutions, and new career development pathways to train future researchers. They also stress that these upgrades are not just about advancing science, they’re necessary for informing coastal planning, resource management, and national climate strategy.
FUTURE 2024 PI-team, Appelgate, B., Dugan, B., Eguchi, N., Fornari, D., Freudenthal, T., et al. (2025). “The FUTURE of the US marine seafloor and subseafloor sampling capabilities.” AGU Advances, 6, e2024AV001560. https://doi.org/10.1029/2024AV001560
Corresponding author: Masako Tominaga, mtominaga@whoi.edu.
FABLE: A Localized, Targeted Adversarial Attack on Weather Forecasting Models
Deng et al. (2025) present FABLE, a new framework that tests how small, targeted changes to weather data can disrupt AI-based forecasting models. Their research shows that minor tweaks to inputs, when focused on specific locations and short periods of time, can lead to big errors in predicted weather.
The researchers evaluated FABLE using a simple simulation and FourCastNet, a leading AI weather model. In both cases, the framework introduced small changes that escaped standard data checks but reduced forecast accuracy. This highlights a key concern: modern forecasting systems can be highly sensitive to limited disruptions, even when those disruptions appear minor or go unnoticed.
The study concludes that as forecasting moves away from traditional physics-based models and toward machine learning, these systems could be more vulnerable to manipulation or failure when the data isn’t perfect. The researchers suggest that to make weather prediction models more trustworthy, they need to be tested for how well they hold up under targeted stress. FABLE raises new concerns about the reliability and security of AI weather systems, especially as they’re used in disaster planning, farming, and transportation.
Deng, Y., Galib, A. H., Lan, X., Tan, P.-N., & Luo, L. (2025). “FABLE: A localized, targeted adversarial attack on weather forecasting models.” arXiv. https://arxiv.org/abs/2505.12167
Corresponding author: Yue Deng, dengyue1@msu.edu.