Science Project Summaries

Gaia is a cornerstone ESA mission that has observed over 1 billion stars. On Dec 2, 2026, it will release its 4th dataset, containing 66 months of data on all these stars. Over the course of the semester, we will learn to mine this rich, stupendously large dataset for gems and discover the colorful universe contained within. Each star can reveal something special and advance our understanding of the evolving universe further; now you can take an active part in this exploration!

Muscle tissue must repeatedly repair damage throughout an animal’s lifetime, yet most studies of healing examine the response to a single injury. This raises an important question: does tissue respond differently when it is injured repeatedly? Repeated injury could enhance healing by priming cells and signaling pathways involved in repair, or it could impair restoration if repeated damage depletes cellular resources or disrupts tissue organization. Understanding how tissues respond to repeated injury may therefore provide insight into how regenerative capacity is maintained over time.

This project will use the jellyfish Clytia hemisphaerica (Clytia) to investigate how repeated injury affects muscle tissue repair. Clytia provides a simple and experimentally tractable system in which muscle-like contractile tissues can be injured and monitored in living animals. Using live imaging, we will compare the response to an initial muscle injury with the response to subsequent injuries delivered after the tissue has had time to repair. We will examine wound closure dynamics, including the rate and extent of wound closure and changes in tissue organization.

By comparing animals experiencing a single injury with animals subjected to repeated injuries, we will determine whether previous injury alters the repair response. We will also ask whether the effects of repeated injury depend on the interval between injuries, providing insight into whether tissue requires sufficient time to restore its regenerative capacity.

This work will establish a foundation for understanding how muscle tissue responds to repeated damage. More broadly, it will test whether repair is simply a response to each individual injury or whether tissues can be altered by their history of injury. The findings may reveal previously unrecognized mechanisms that regulate regenerative capacity and tissue resilience.

The proteasome is a macromolecular machine that unfolds and degrades proteins in all eukaryotic cells. In the Kraut lab, we study the processivity of the proteasome – that is, its ability to unfold and degrade substrates containing folding domains without falling off the substrate. Substrates to be degraded are polyubiquitinated, meaning a chain of small proteins called ubiquitin is attached to lysines within the substrate. We are currently in the process of establishing an unfolding ability assay within yeast to measure proteasomal processivity. Substrates consist of a protein containing a degron region that is ubiquitinated, followed by two fluorescent proteins: one easy to unfold (either LSSmOrange, an orange fluorescent protein, or IRFP670nano3, an infrared fluorescent protein) and one difficult to unfold (GFP, green fluorescent protein). The LSSmOrange or IRFP670nano3 is easily degraded by the proteasome, but the GFP sometimes causes the proteasome to stall, and thus the ratio of fluorescence values between the two proteins reports on the extent to which the proteasome stalls on GFP. However, preliminary experiments with purified proteins indicate that GFP-containing fragments may be more fluorescent than GFP in the full-length substrate containing two fluorescent proteins. This project aims to investigate the fluorescent properties of full-length versus truncated substrates by purifying proteins containing a defined cleavage site between the two fluorescent proteins. Results can be used to calibrate fluorescence values from in vivo assays to give more quantitative information about unfolding abilities.

Anticancer peptides are short amino acid sequences whose biological effects may depend on molecular properties such as charge, hydrophobicity, amphiphilicity, three-dimensional structure, and conformational flexibility. Experimental studies have identified peptides and sequence variants with different levels of activity and selectivity, but the molecular consequences of these sequence differences are not always predictable.

This project will use molecular dynamics simulations to investigate how defined sequence differences and selected mutations influence the structure and conformational behavior of experimentally characterized anticancer peptides. A small candidate set will be selected from systems with suitable structural information or tractable starting models, potentially including pleurocidin-family peptides such as NRC-03 and NRC-07, the short peptide Temporin-SHf, or the membrane-active peptide LTX-315. The student will first reproduce a validated simulation workflow for one lead peptide and characterize its conformational ensemble in water. A closely related peptide or experimentally motivated sequence variant will provide an initial comparison.

After establishing the reference behavior, the student will investigate carefully selected amino acid substitutions designed to modify a specific property, such as net charge, hydrophobicity, amphiphilicity, or helical propensity. Multiple independent simulations will be conducted for the parent peptide and selected variants. The resulting trajectories will be compared using observables such as flexibility, radius of gyration, hydrogen bonding, secondary-structure content, solvent exposure, and conformational clustering. Sequence-based properties, including net charge, hydrophobicity, and hydrophobic moment, will help connect chemical changes with molecular behavior.

The project will identify reproducible and physically interpretable relationships among peptide sequence, structure, and dynamics. It will produce a validated computational workflow, comparative molecular-level insights, and a rationale for prioritizing peptide variants. The resulting systems and hypotheses will provide a foundation for examining peptide interactions with compositionally distinct membrane models, supporting broader research on membrane selectivity and the molecular design of peptide therapeutics.

In 2022, I designed the RadioLand mobile app project at Villanova as a graduate student. I continue to maintain it and use it as a teaching tool in my classes. The designated research candidate will assist me with the installation of Python-based workflows to keep sports affiliates, NPR affiliates, and streaming URL listings within the app up to date. This will broaden their programming and data analysis knowledge and immediately introduce them to the use of third-party APIs. A Computer Science or Computer Engineering major is ideal, but not mandatory.    

Small networked devices like cameras, sensors, and smart-home hardware are attacked constantly, but they are too weak to run modern intrusion-detection models themselves. The obvious fix is to ship their traffic to an edge server, except that it uses bandwidth and battery the devices don't have. My research asks a different question: instead of sending raw data, can a device learn to send only the meaning the detector actually needs?

Our system pairs two learned components. A neural ""semantic encoder"" compresses a network flow into a handful of numbers chosen for detection, not reconstruction, so most of the data is never transmitted. A reinforcement-learning agent then decides, per flow and in real time, how aggressively to compress and whether to process locally or offload, trading accuracy against latency, bandwidth, and energy as channel conditions change. 

The student will build on this in two ways:
First, they will train the compressor so that its output is ranked by importance and the most useful information comes first. That way, one trained model can send just two numbers when the connection is bad and eight when it is good, instead of us having to train and store a separate model for every setting. 

Second, they will build a testing setup that pushes the system into situations it never saw during training: a slow network, a busy server, or far more (or far fewer) attacks than usual. They will measure how well it holds up using metrics that stay honest when attacks are rare, and repeat each test with several random starting points so the results are trustworthy rather than lucky.

Automated vehicles keep their distance using a simple rule: measure the gap to the car ahead and its relative speed, then accelerate or brake accordingly. Written down, this rule is a differential equation, and a line of such vehicles is a chain of coupled differential equations whose behavior is well understood.

This project asks what happens when the measurement is a lie. An attacker who has compromised one vehicle does not need to seize the steering wheel. It is enough to shift the reported gap by a small amount. The student will add such a false-data term to a standard car-following model and answer two questions:

•  First, how small can the lie be and still destabilize the line? This is a classical stability question with a closed-form answer that the student can derive by linearizing the model, then confirm numerically.
•  Second, how small can the lie be and still go unnoticed? The student will build a simple monitor that watches the difference between predicted and observed behavior, and find the smallest deviation it can flag at a fixed false-alarm rate.

Generative AI can write code, explain code, find bugs, and suggest solutions in seconds. When an AI assistant helps you complete a programming assignment, however, an interesting question arises: How do you know whether you learned how to solve the problem—or the AI simply helped you finish it?

This project explores an increasingly important question in computer science education: What does it mean to understand programming when AI can produce the code? A program that produces the correct output tells us that the solution works, but it may not tell us everything about what the student understands. Can the student explain the idea behind the solution? Can they describe the algorithm without relying on programming-language syntax? Can they reason about why the program works or apply the same idea to a different problem?

We will investigate different ways students can demonstrate their understanding of a programming problem, including working code, pseudocode, explanations of their solutions, and conversations about how and why their solutions work. We are particularly interested in whether these different forms reveal aspects of students' problem-solving and algorithmic thinking that may not be visible from correct code alone. We will also explore how generative AI can be used to ask follow-up questions during these conversations and help us better understand what a student knows.

Language models have made significant advances in recent years.  Their applications can be seen in areas ranging from autocorrect systems to modern AI agents.  This project aims to develop an AI teaching assistant for an introductory computer science course.  This will then be deployed on a humanoid robot as an example of embodied AI use.

The MATCH student will investigate techniques for developing efficient AI models using various methods, which may include model distillation, teacher-student training, and retrieval-augmented generation.  Model responses will be evaluated for accuracy, speed, and clarity.  A primary constraint for the model will be that it must run within the humanoid robot’s hardware. The humanoid robot has a microphone and speaker for auditory interactions, which differentiates the AI system from more common text-based LLM interactions.

The model will be specifically tailored to the content from an introductory computer science course to act as a teaching assistant.  The project will explore the challenges of deploying language models on embodied platforms, such as humanoid robots.  This project may lead to future evaluation of the effectiveness of text-based LLMs compared to embodied language models on a humanoid robot in aiding student comprehension.

The trade of reptiles for exotic pets is an expansive industry that can threaten biodiversity. While much of this trade is both legal and sustainable, it can simultaneously contribute to species declines by incentivizing overexploitation of wild populations, facilitating the spread of invasive species, and acting as a vector for the transmission of zoonotic pathogens. In August 2026, a new documentary series called Monsters of God was released on the streaming platform HBO Max, highlighting the history of illegal and unsustainable reptile trade in the United States. These high-profile documentaries have significant potential to shift the regulatory and commercial landscape of wildlife trade; for example, the docu-series Tiger King and Chimp Crazy, created by the same filmmaker as Monsters of God, are recognized as having directly spurred both federal and state-level regulations restricting the trade in exotic mammals in the United States.

The Weston Lab works on questions around the resilience of tidal wetland ecosystems in the face of climate change and land use change. Tidal wetlands provide many critical ecosystem services, but many wetland systems are threatened by sea-level rise, changing inputs from the watershed, and other impacts of human activities. The First Year Match students will work with the Weston lab on research in support of several externally funded projects that examine carbon and nitrogen cycling in coastal wetland ecosystems and the response of coastal wetlands to sea-level rise and changing land use. The laboratory conducts fieldwork in the Spring, Summer, and Fall in Plum Island Sound, MA, and more locally in the Delaware River estuary. Samples collected in the field are brought to the laboratory for processing.

The recent influx of hydraulic fracturing, or “fracking”, for natural gas production in Pennsylvania has prompted societal concerns about its potential impacts on human health and the environment. Currently, the extent to which the proximity and density of hydraulic fracturing well sites affect local water quality remains unresolved. This study utilizes historically collected streamwater and riverbed sediments to determine the relative impacts of nearby hydraulic fracturing well pads on stream water quality. Streamwater and riverbed sediments were collected from 47 headwater tributaries of the Upper Susquehanna Watershed in northern Pennsylvania, an area of high drilling activity, from 2015 through 2018. Students will help analyze water samples for major elements and trace metals, and streambed sediments for the latter. The data will then be used to determine whether a statistical relationship exists between the number of fracking wells and/or violations in a watershed and the associated streamwater/riverbed sediment geochemistry. The study results are expected to help regulators and decision-makers overseeing fracking activity in their local communities.

Coastal wetlands provide habitat for wildlife and serve as a buffer against extreme flooding. Human activity, exacerbated by rising sea levels, threatens the health and longevity of this critical ecosystem. Prime Hook National Wildlife Refuge, Delaware, USA, was managed as an artificial freshwater habitat until Hurricane Sandy breached the shoreline. A subsequent restoration project aimed to bring salt marshes back to Prime Hook. This project will use geologic techniques to explore restoration progress and investigate the long-term history of salt marsh health.

Mathematical modeling is a useful technique for describing how systems change over time and testing hypotheses that may be difficult or impossible to explore in reality. This project investigates the transmission of Clostridioides difficile, an infection-causing bacterium that was once largely associated with healthcare settings but is now increasingly acquired in the community. In some regions, community-associated C. difficile infection (CA-CDI) accounts for nearly half of all reported cases, highlighting the need to better understand the mechanisms driving community transmission.

C. difficile subsists in the environment as spores that can survive for long periods. Individuals can become colonized (asymptomatically infected) after exposure to contaminated surfaces or infected individuals. However, exposure does not necessarily result in infection. Factors that disrupt the gut microbiome, such as recent antibiotic use or increased age, can increase susceptibility. Asymptomatic carriers can also spread C. difficile without experiencing symptoms, making them an important part of transmission. 

We will develop a system of ordinary differential equations (ODEs) to investigate CA-CDI transmission. The model will divide the population into three age groups (infants, adults, and elderly individuals) and track changes in disease states over time. This model will incorporate key factors involved in transmission, including antibiotic use, environmental contamination, and person-to-person transmission. Simulations will be used to investigate mechanisms sustaining CA-CDI and evaluate potential interventions, including improved antibiotic stewardship and enhanced environmental cleaning. Results can provide insight into the mechanisms sustaining CA-CDI and offer guidance for public health strategies to reduce the growing burden of C. difficile. 

The student researcher will help formulate, analyze, and simulate the model and evaluate how changes in model parameters affect CA-CDI transmission. More broadly, the project will demonstrate how mathematical models can be used to investigate complex infectious disease systems, test hypotheses, and evaluate potential strategies to inform public health decision-making.

Graphs are mathematical models of networks, consisting of vertices joined by edges. An arithmetical structure assigns positive integers to the vertices according to divisibility rules: each vertex’s label must divide the sum of its neighbors’ labels, and the labels must have no common factor greater than one. These objects connect graph theory, number theory, and linear algebra. A central research question asks how the arrangement of edges influences the number of arithmetical structures a graph admits.

This project will combine mathematical exploration and computer programming to develop tools for studying that question. The starting point is an algorithm developed by Valencia and Villagrán for enumerating arithmetical structures. We will study the algorithm and its existing implementation, then investigate how to adapt the computational workflow to accept a single specified graph as input. This would allow targeted computations without processing all graphs of the same size and make it easier to investigate particular examples and families of graphs.

Using the adapted code, we will build a database recording graphs, their arithmetical structures, and the corresponding counts. Our initial target is connected simple graphs with up to six vertices, with the scope adjusted according to computational feasibility. We will check results against known examples and distinguish completed enumerations from computations still in progress. The resulting data will help identify patterns, test conjectures about which graphs admit the most or fewest structures, and guide further mathematical investigation.

Through guided reading, small examples, and progressively more independent computational work, the student will contribute to an ongoing research program while developing skills in mathematical reasoning, programming, and organizing reproducible research. Expected outcomes include documented code and a carefully checked database that can be expanded and used in future research.

Atomic Force Microscopy (AFM) has emerged as an indispensable tool in the study of low-dimensional materials of both hard and soft matter.  In this study, we use AFM to study human ferritin heteropolymers overexpressed in E. coli bacteria by genetic engineering techniques. Ferritin is the iron storage protein in living systems where iron, an essential element for life, is sequestered within a spherical protein shell of 8nm inner and 12nm outer diameter, composed of two types of amino-acid chains, H and L. These two types of chains play complementary roles in iron reduction, nucleation, and nanoparticle growth within the shell. H-rich ferritins are associated with heart and brain tissue, where fast iron trafficking occurs, while L-rich ferritins are associated with liver and lung tissue, for long-term iron storage. Ferritin malfunction is associated with a host of iron-related disorders, including neurological disorders.  Thus, ferritin is of great interest to the physiology of iron homeostasis in health and disease. In addition to ferrihydrite, other metal and metal oxide nanoparticles can be grown within the protein shell, which presents itself as a robust nanotemplate to produce monodisperse nanoparticles, of interest to nanoscience and nanoengineering (nano-biomedicine, targeted drug delivery, nano-architectural designs for device applications, etc.). 

The nanomechanical properties of heteropolymeric shells will be studied by investigating the elasticity, deformation and Young’s modulus of H-rich and L-rich heteropolymers in the absence and presence of a ferrihydrite core. The student will be trained in the preparation of biological samples for AFM investigations and the operation of this diverse, multimode microscope to obtain topographical maps of ferritins and measure the elastic properties of various ferritin samples.  This study investigates the Physical properties of soft, biological matter and is of relevance to biology, bioengineering and nanotechnology.

Chemical reactions are usually determined by the properties of molecules themselves. However, recent experiments have shown that reactions can change when many molecules are placed inside a tiny optical cavity, a device that traps light in a very small space. In these systems, light and molecular vibrations can interact strongly, creating a new environment that may influence how molecules store energy and how chemical bonds break. Surprisingly, these changes can occur even when the light is not perfectly matched to the molecular vibrations, indicating that important questions remain unanswered.

This project aims to understand how light trapped in an optical cavity affects the breaking of molecular bonds. Using computer simulations, we will model collections of molecules and follow how their vibrational energy changes over time. We will investigate whether molecules break apart more quickly or more slowly when they interact collectively with the cavity field and how this behavior depends on the number of molecules present.

A key question is whether the molecules all need to start in the same state or whether differences between molecules change the outcome. To answer this, we will compare simple situations, where all molecules begin identically, with more realistic cases where molecules start with different energies, similar to what happens at finite temperature. We will also explore whether the behavior of many interacting molecules can sometimes be explained using a simpler single-molecule picture.

The project will use existing simulation tools developed by the Rokaj research group and will be carried out in Wolfram Mathematica. This work will give first-year undergraduate students hands-on experience with scientific computing, data analysis, and modern research while introducing them to exciting questions at the intersection of physics, chemistry, and light-matter interactions.

A supernova is a catastrophic explosion at the end of the life of some stars. There are several types of supernovae, each the result of unique physical and chemical processes during the star's collapse. Many of these differences can be distinguished using spectroscopy, a process that divides electromagnetic radiation into a spectrum, but it can be expensive and time-consuming. Photometric data (measuring the brightness in one part of the spectrum) is much more widely available, however, especially with the arrival of the next generation of instruments such as the Rubin Observatory. Time-domain surveys like the Legacy Survey of Space and Time take repeated images of the sky every few nights, giving ample data to study billions of objects, including thousands of supernovae every night!

Upcoming telescope receivers designed to measure the cosmic microwave background (CMB) and emission lines from faint, unresolved galaxies (LIM) are poised to advance our understanding of cosmic inflation, dark energy, and beyond-Standard-Model physics. In building these receivers (e.g., SPT-3G+), spline-profiled feedhorns are a compelling technology for optically coupling light from telescope-optics systems to the superconducting-detector arrays that will be used for these measurements. While spline-profiled feedhorns offer a desirable balance between coupling efficiency and beam-polarization effects, they are relatively challenging and expensive to fabricate. Their cost and feasibility scale with the machining tolerances required by the designs; however, the exact tolerancing needs are not well known.

We propose investigating the optical performance of spline-profiled feedhorns with varying types and degrees of fabrication errors. To do so, we will use simulation software (e.g., Ansys HFSS) to model the electromagnetic performance and beam properties of several different horns, each with physically realistic errors in varying degrees. The performance of these horns will be compared to the simulated performance of an ideal spline-profiled horn with the goal of understanding and quantifying how relaxing machining-tolerance requirements affects the optical performance of these components. This information will be used to inform the design of next-generation experiments, potentially yielding significant savings in cost, time, and complexity.

While we anticipate a good deal of work will be accomplished over the course of the spring semester, if the student is interested, the possibility of securing funding to extend this project through the summer is very high. At the end of this investigation, it is our strong intent to submit this work for publication in a refereed journal (e.g., Applied Optics).

Lesbian, gay, bisexual, transgender, queer, or otherwise marginalized based on their sexual orientation or gender identity (LGBTQ+) individuals consistently exhibit higher rates of mental health problems compared to their heterosexual, cisgender counterparts. These disparities stem from a variety of sources, including proximal and distal minority stressors, as well as intracommunity stressors. One mental health domain of particular concern to researchers is negative body image and eating disorders, areas in which LGBTQ+ individuals consistently report elevated rates.

One potentially promising avenue for preventing the development of negative body image in this population is participation in LGBTQ+ community-focused sports. Across groups and cultures, sport participation offers numerous physical and mental health benefits. Research consistently demonstrates that LGBTQ+ individuals experience unique benefits from participation in LGBTQ+ community-focused sporting leagues (for a review, see Scheadler, 2026).

However, the literature examining LGBTQ+ sport participation and body image remains limited. This study draws on a sample of 250 LGBTQ+ individuals who participate in queer sports leagues, with data collection beginning this fall. Participants will complete a survey consisting of quantitative measures and open-ended questions designed to capture their experiences in LGBTQ+ community-focused sporting leagues and to better understand the impact of such participation on body image and broader well-being.

For the purposes of this project, we will analyze the open-ended responses. Specifically, we will use Spangler et al.'s (2012) consensual qualitative research-modified approach, an inductive qualitative analytic method, to examine how LGBTQ+ community members perceive sport participation to have shaped their experiences of their bodies.

The lab has been studying the effects of early life stress on susceptibility to subsequent stressors for years. We are currently testing the hypothesis that the long-term consequences of early life stress depend both on stressor intensity and duration and on the exact developmental period in which stress is experienced. This Match project focuses on the neurobiological consequences of stress and will hopefully provide insight into the basis of behavioral consequences of stress that are being examined in parallel work.

Mosquitoes are vectors for transmitting several human and animal parasites. Once ingested by the vector, parasites must avoid being eliminated. One strategy employed by kinetoplastids, a group of single-celled parasites, is to physically attach to the epithelial lining of the vector's digestive system. The specific site along the gut where parasites attach varies among different kinetoplastid species, but attachment itself is a conserved and essential process. We are interested in uncovering the molecular and cellular mechanisms governing parasite attachment. Studying this process in vivo is challenging because kinetoplastids are tiny, measuring only about 5 µm in diameter. A revolutionary technique called expansion microscopy (ExM) promises to transform how we study this interface. Instead of relying on expensive super-resolution optics, ExM physically increases the sample's size using an expandable polymer gel. By literally blowing the tissue up like a balloon, nanoscale interactions can be clearly resolved using standard fluorescence microscopes.