2026 Undergraduate Summer Internship Program information
Project 1: Machine Learning for Predicting Properties of Pathogens
Host: Dr. T. M. Murali — Virginia Tech – murali@cs.vt.edu
Project Description: This project will seek to answer one of the fundamental mysteries of pandemic prediction: what factors cause a pathogen to shift host organisms? Researchers in the COMPASS Center are developing machine learning models to address this question. Current approaches can predict the host organism that a virus may infect, purely from the sequence of a viral protein. Specific goals of this project will be to extend these methods to predict other properties, to evaluate them in novel ways on unseen data, or to train them using different types of data.
The selected student will join a thriving and collaborative multi-institutional and multi-disciplinary research team.
They will learn how to:
- Develop new machine learning and graph algorithms to solve biomedical problems, specifically in the context of pandemic prediction and prevention.
- Obtain a deep understanding of different approaches by quantitative and rigorous comparisons.
- Conduct high-quality experiments to evaluate cutting-edge research ideas implemented in software.
- Contribute effectively to impactful, open-source software development.
Desired skills/background/expertise: Substantial experience in Python. Knowledge of machine learning packages such as PyTorch will be very useful.
Project 2: The influenza virus polymerase as a determinant of host range and adaptation
Host: Dr. Adam Lauring — University of Michigan – alauring@med.umich.edu
Project Description: Influenza viruses are ubiquitous in nature, and the evolution of human influenza A virus (IAV) is characterized by zoonotic spillover events followed by adaptation of these animal viruses to human hosts. Mutations in the polymerase are critical to IAV adaptation, as they determine how well the virus will replicate in its new host. For example, avian influenza viruses often require the PB2 E627K substitution in order to replicate efficiently in mammalian cells, and this mutation has repeatedly been observed in experimental virus evolution in laboratory cell lines and animals as well as in viruses recovered from humans infected with 1918 H1N1, H5N1, and H7N9 IAV. The polymerase also determines influenza virus’ mutation rate, and therefore the rate at which it will acquire mutations that lead to host range expansion, drug resistance, or antigenic drift.
We are broadly interested in understanding the short-term and long-term evolution of the influenza virus polymerase and its relationship to disease emergence. We have developed assays to precisely measure the fitness of influenza viruses in culture and their mutation rates under a range of conditions. We are also broadening our work on deep mutational scanning of the polymerase. Here, we generate libraries that contain every possible point mutation in a polymerase protein and measure the impact of each in a high throughput screen. Summer students will gain exposure to all of these assays and an appreciation for approaches to study evolution and disease emergence. This work has applications to understanding the basic biology of influenza viruses as well as antiviral approaches.
Project 3: Development of an microphysiological chip incorporating an immune system
Host: Dr. Colin Bishop — Wake Forest Institute for Regenerative Medicine – Colin.bishop@wfusm.edu
Project Description: Developing an integrated chip that incorporates a functional human immune system would significantly enhance the physiological relevance of in vitro disease and drug-testing models. Immune responses play a central role in infection, inflammation, tissue injury, and repair, yet most current organ-on-chip platforms lack dynamic immune components and therefore miss critical host–pathogen and host–tissue interactions. An immune-integrated chip would allow precise control over immune cell recruitment, activation, and signaling, enabling mechanistic studies of innate and adaptive immune responses in a human-relevant context. This capability is especially important for understanding immune-mediated pathology, immune evasion by pathogens, and variability in patient responses that cannot be accurately captured using static cultures or animal models.
Such a platform would also be transformative for translational and therapeutic development. By modeling immune–tissue crosstalk in real time, an immune-enabled chip could improve prediction of drug efficacy, toxicity, and immunogenicity, including cytokine-mediated side effects and off-target immune activation. This system would support evaluation of vaccines, biologics, and immunomodulatory therapies under controlled, reproducible conditions, while reducing reliance on animal testing and early-stage clinical risk. Overall, integrating immune functionality into microphysiological systems would accelerate discovery, de-risk development pipelines, and provide a more predictive framework for studying complex human diseases.
Desired skills/background/expertise: Tissue culture expertise
Project 4: Understanding Public Perceptions of AI in Pandemic Prediction
Host: Dr. Julie Gerdes — Virginia Tech (Arlington Research Center) – jgerdes@vt.edu
Project Description: As artificial intelligence tools take hold in the public consciousness through its deployment in classrooms and workplaces across sectors, levels of acceptance have varied widely. For example, a recent internal study at Virginia Tech found that undergraduate students in writing classes perceived that the risks of inaccurate returns, plagiarism accusations, and negative impact on their education to be greater than the benefits of genAI tools. Still, other reports indicate widespread use of the tools in writing tasks, including in higher education. It is unclear how these sentiments transfer to the perceived ethics of using data from animal hosts (e.g. pigs, humans, and other mammalian and bird species) to anticipate pathways for viruses not previously studied. For pandemic prediction research, public perception of the use of large language models trained on viral protein sequences could influence data access and availability, funding opportunities, and data governance. Additionally, we predict that perception varies based on access to scientific messengers, geographic location, religion, mistrust in public health authorities due to historic marginalization on the basis of race or economic status, and economic reliance on work at the human-animal interface. This survey study will analyze public perceptions of artificial intelligence for virus-host prediction using an experimental approach. Results will lead to recommendations for data use and sharing policies that reflect public concerns as well as tailored communication strategies about predictive science in pandemic research.
The student intern will have the opportunity to provide input on study design and will work directly with Dr. Gerdes to analyze survey results. Primary responsibilities will include reviewing literature and conducting statistical analysis of quantitative results. The student will develop skills in academic writing, collaborative research, and science communication. They will have learn about pandemic science, health equity, and risk communication. The selected intern will walk away with writing samples in the form of a research report and brief and/or a coauthored academic article. This student will be based at the Arlington Research Center in Northern Virginia at least two days a week but will interact with other members of the NSF COMPASS Center during routine remote meetings.
Desired skills/background/expertise: This student should be an upperclassman (senior preferred) in public health, statistics, computer science, communication, bioethics, or a related field. They must have experience analyzing survey data in SPSS and/or R and/or completed courses in statistics or quantitative research. They should possess an interest in the ethics of pandemic research and/or pandemic risk communication, particularly in social perceptions of the use of machine learning/AI with a focus on health equity.
Project 5: What’s in a Vector? The role of ticks in mosquito-borne virus spillover
Host: Dr. Laura Goodman — Cornell University, Baker Institute for Animal Health – laura.goodman@cornell.edu
Project Description: The Goodman lab’s role in the COMPASS center is on Use Inspired Research Project 3 – Defining the suite of West Nile Virus (WNV)-competent vectors in North America. A “competent” vector has the ability to maintain and transmit a pathogen to a new host. The ultimate goal of this project is to use machine learning models (developed in the Jump thrust) to predict the competence of non-mosquito arthropods, providing a stress test for model performance in data limited systems.This project is a collaboration with Speer Lab at Northern Arizona University, who is focusing on museum archives, while our group is working with prospectively collected tick and animal samples. We also collaborate closely with the Bento Lab at Cornell, the Northeast Center of Excellence for Vectorborne Diseases, and Cornell Cooperative Extension.
We have a large collection of ticks and animal specimens from the Northeast and mid-Atlantic regions of the USA. Our initial targeted screening in NY has pointed to three species, Dermacentor variablis, Amblomma americanum, and Haemaphysalis longicornis as possible spillover vectors. We use mitochondrial DNA analysis to confirm the sources of bloodmeals on pathogen-positive ticks, to either match the sources with known animal reservoirs, or discover unknown reservoirs. We then further characterize the pathogen genomes to compare them to strains circulating in humans.
In addition to studying known viruses, you will also take part in novel virus discovery for pandemic prediction. We recently discovered two novel viruses. First, a putative novel flavivirus in Haemaphysalis longicornis, the invasive longhorned tick. Based on the sequences we have been able to identify thus far, it appears to be the first example of a chimera between segmented and non-segmented flaviviruses. Codon usage analysis indicates a strong adaptation for humans across both components of the genome, followed by dogs and raccoons. You will be a part of a team working on this novel virus in parallel with the WNV-focused experiments, including testing potential host specimens for evidence of previous exposure.
As a summer intern, you will learn how ticks are tested for known and unknown pathogens using cutting-edge molecular methods. You will have the opportunity to contribute to two scientific manuscripts: a review paper on the use of tick bloodmeal remnant analysis for virus ecology studies and a research report on the screening and prioritization of vector species for WNV. You will be working with other students in a variety of degree programs including both veterinary and public health. While you will work primarily at the bench, you may participate in analysis of genetic and geographic data if interested.
Desired skills/background/expertise: We are looking for someone who is detail-oriented, organized, and eager to learn new skills in a teamwork-centered environment. Prior wet lab experience outside of class is a plus but not required.
