2025-2026 Pilot Project Investigators
| April Kloxin, PhD Joint Professor in Materials Science & Engineering & Associate Department Chair Institution: University of Delaware Website Project Title: Antigen Displaying Peptide Nanoparticles Summary: This proposal seeks to develop a novel peptide nanoparticle platform for enhanced B-cell activation by integrating precise control of antigen display with modular nanoparticle design. Synthetic vaccines, especially those using virus-like nanoparticles, are crucial in immunotherapy, but current systems lack the precision necessary for optimal therapeutic use. To address this, the investigator proposes a peptide-based nanoparticle system, which is termed a “bundlemer.” The bundlemer is a robust, cylindrical peptide nanostructure (2x2x4 nm³) that can be functionalized using click chemistry for modular antigen presentation. These epitope-displaying bundlemers (EDBs) offer a transformative approach to creating virus-like nanoparticles, combining precise chemical and structural definition with the ability to control size, rigidity, and multivalent antigen display. The project has two aims: 1) Synthesize and purify bundlemer rods with a narrow size distribution and 2) create bundlemer rods with controlled antigen spacing and presentation. Functionalized bundlemers will display peptide epitopes at predetermined intervals, enabling the study of size, spacing, and density effects on B-cell activation. The research team will conjugate peptides with known B-cell receptor affinities and assess synergistic effects. This pilot work will establish a versatile platform for antigen presentation, enabling nanoscale control over immune activation. The bundlemer construct offers a generalizable tool for vaccine development and immunotherapy, with broad potential applications in biomedical research. |  |
| Jeremy Bird, PhD Assistant Professor Biology Institution: University of Delaware Website Project Title: Systematic Molecular Characterization of Mitochondrial Transcription Initiation with both ATP and NAD Summary: Regulation of gene expression is essential to the proper function of a cell and an organism. The first step in gene expression, transcription, is also the most highly regulated. This regulation involves not only transcription factors that affect RNAP activity on a gene, but also the enzymatic functions of RNAP itself including affinity for promoter sequence and the DNA sequence-dependent rate of synthesizing new DNA. A thorough grasp of the mechanism of RNAP in transcribing DNA and the regulation of this process is key to fully comprehending how a cell operates. Mitochondria genomes encode their own complement of ribosomal RNAs, tRNAs and essential protein subunits of the oxidative phosphorylation pathway. These genes are transcribed by a mitochondrial-specific single subunit RNA polymerase (RNAP) in a process regulated by mitochondrial-specific factors. Expression of the mitochondrial translation machinery and oxidative phosphorylation subunits must be carefully coordinated with expression of nuclear encoded mitochondrial proteins to ensure proper function. It has been hypothesized that the rate of mitochondrial transcription is directly regulated by the availability of the initiating nucleotide Adenosine Triphosphate (ATP) based on thr Km of ATP binding to transcription complexes during transcription initiation at the different mitochondrial promoters. However, we have demonstrated that mitochondrial RNA polymerase also use Nicotinamide adenosine diphosphate (NAD) as a Non-Canonical Initiating Nucleotide (NCIN). Transcripts from both S. cerevisiae and human Mitochondria carry 5´ NAD caps. We hypothesize that regulation of mitochondrial transcription is governed by the availability of both ATP and NAD as initiating nucleotides. Importantly, the use of NAD as an NCIN by mitochondrial RNAPs potentially serve to regulate mitochondrial transcription in response to the metabolic state of the cell by influencing the kinetics of transcription initiation based on the concentration of both ATP and NAD available to RNAPs. To address our hypothesis, we propose to do the following: 1. Using a global in vitro transcription assay with long read RNA-sequencing as output to comprehensively define the transcriptional units of the S. cerevisiae mitochondrial genome. Establish the extent of NAD-capping of mitochondrial transcripts from tissues with varying metabolic programs to determine the link between transcription initiation and cellular metabolic state. |  |
| Shanshan Ding, PhD Associate Professor of Statistics Assistant Professor Institution: University of Delaware Website Project Title: New Statistical Inference and Learning for Understanding Cellular Heterogeneity in Single Cell Genomic Data Summary: Single-cell RNA sequencing (scRNA-seq) is a cutting-edge technique that allows researchers to measure the gene expression profiles of individual cells. Driven by the importance of scRNA-seq data, various statistical and machine learning methods have been designed for analytics tasks including cell clustering, imputation, differential expression analysis, gene network inference, among others. Classical differential expression analysis typically performs testing for expression differences on individual genes separately and often overlooks the potential interactions between genes. Ignoring these interactions can lead to less efficient use of the data and incomplete biological interpretation of the data that miss to identify key genes. In addition, traditional scRNA-seq methods do not capture the full spectrum of gene expression variability and the potential multi-modal nature of gene expression distributions. In recent years, various self-supervised deep generative modeling (DGM) methods have been developed to model complex, non-linear distributions of gene expression data. As gene regulatory and co-expression networks/graphs provide valuable prior knowledge, integrating this prior network information into DGM holds the promise of learning improved latent features and representations based on biologically meaningful aggregation. The overarching goal of this proposal is to develop novel scRNA-seq data analysis methods for tackling aforementioned new challenges encountered in differential expression analysis and DGM modeling. These new techniques can also lead to more effective analysis in cell clustering and can be extended to single-cell multiomics data. |  |
2025-2026 Research Project Investigators
| Jennifer K. Peterson, PhD Assistant Professor / Medical Entomologist Institution: University of Delaware Website Project Title: Harnessing an optimized multi-scale, One Health approach to produce evidence-based, actionable estimates of vector-borne disease risk in Delaware Summary: Chagas disease is a vector-borne parasitic infection that can lead to serious cardiac or gastrointestinal alterations that can be fatal. The disease is caused by the protozoan parasite Trypanosoma cruzi and spread by insect vectors known as ‘triatomine bugs.’ The most widespread triatomine bug species in the United States is Triatoma sanguisuga. Despite its public health importance, T. sanguisuga has been studied in only a small portion of its geographic range, which spans from the Rocky Mountains to the eastern seaboard. In 2023, I discovered the first T. sanguisuga infected with T. cruzi in Delaware. I then initiated the first investigation of T. sanguisuga in the northeastern US, funded by a 2024-25 INBRE pilot award. In the INBRE supported pilot project, we took the first step toward establishing the importance of Trypanosoma cruzi and its vector in Delaware. We gathered preliminary data on vector and parasite occurrence, T. cruzi infection in the vector, and human blood meal incidence in the vector. To date we have collected 55 T. sanguisuga specimens distributed across human homes, schools, businesses, and recreational areas. We found a T. cruzi prevalence of 13.5% in all T. sanguisuga specimens collected, and a prevalence of 23.5% in adult specimens. Nine specimens from four sites had human blood in them. Three specimens, found at two human homes and a school, were infected with T. cruzi AND had human blood in them. Subsequent DNA-based analyses with higher sensitivity will likely reveal a higher T. cruzi prevalence as well as the animal species on which the bugs are feeding. Nevertheless, our findings convey a critical message: T. cruzi-infected T. sanguisuga are feeding on humans in Delaware. There is a clear need and justification for the expansion of this human health research that will build upon our successful first year. In the proposed project, I will use our pilot project findings to optimize and expand our T. sanguisuga sampling effort throughout Delaware. I will then use the field data collected in both 2024 and 2025 to develop, train, and validate predictive ecological and epidemiological models that will be used to identify areas of highest Chagas disease exposure risk in Delaware, which will be visualized in risk maps. Findings from this research will ultimately empower citizens to make evidence-based decisions when encountering a Chagas disease vector in their home, school, or place of business, thus increasing public health preparedness in Delaware. |  |
| M. Shahidul Islam, PhD Associate Professor, Department of Chemistry Institution: Delaware State University Website Project Title: Computational-Experimental Screening of Peptides for Triple Negative Breast Cancer Therapy Summary: Triple-negative breast cancer (TNBC) is a very aggressive subtype of breast cancer, which has a poor prognosis, drug resistance, and high malignancy. Approximately 15-20% of breast cancer cases are diagnosed as TNBC, affecting primarily younger women, and inducing significant emotional and mental challenges. TNBC is negative for three biomarkers, the estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2). The lack of these biomarkers precludes the use of targeted therapies, contributing to TNBC’s higher rates of metastasis and poorer survival outcomes compared to other breast cancer types. Current therapeutics that mitigate malignant transformation and metastases in TNBC primarily include chemotherapy, radiation, and surgery. However, these treatments often induce serious cellular cytotoxicity and adverse effects that lead to poor therapeutic outcomes. Using cutting-edge computational as well as experimental tools, this project aims to generate a high-affinity peptide library targeting mesothelin, a protein highly over expressed in TNBC cells and its expression is almost negligible in normal tissues. In this proposal, we will focus on developing the high-affinity peptide library by combining computational modeling and AI-based drug discovery methods. Included are RFdiffusion, ProteinMPNN, and AlphaFold models, molecular dynamic simulation, experimental biochemistry, and molecular biology techniques such as protein expression and purification (e.g. ELISA, Western Blot, and SDS-PAGE, Biolayer interferometry and Flow cytometry). Successful binding of the designed peptides with mesothelin and subsequent recruitment of natural killer (NK) cells could serve as an effective method to inhibit mesothelin and potentially combat the TNBC. Additionally, peptide-drug conjugates may offer a targeted approach to deliver small molecule drugs, such as cell cycle inhibitors, directly to TNBC cells, leading to their destruction. These peptides could also serve as biomarkers to identify aggressive, high-risk TNBC cases. This project will foster technology development and provide unique training opportunities for talented undergraduate and graduate students at Delaware State University. |  |
| Yixiang Deng, PhD Assistant Professor, Computer & Information Sciences Institution: University of Delaware Website Project Title: Next-generation Glucose Prediction and Control for Diabetes via Scientific Machine Learning Summary: Type 1 diabetes (T1D) affects approximately 2 million Americans, including 304,000 children and adolescents, and requires precise glucose management to prevent life-threatening complications such as cardiovascular diseases and diabetic coma. The ultimate goal of diabetes management is to develop a smart “artificial pancreas” that automates insulin delivery using continuous glucose monitors and insulin pumps. However, achieving accurate and safe glucose control remains a challenge due to poor data quality, limited predictive precision, and the lack of robust, cost-effective control algorithms. This project seeks to advance glucose prediction and control through a novel computational framework that integrates multi-source data, physiological modeling, and advanced machine learning (ML) techniques. Building on our preliminary work in patient-specific digital twins for T1D, we propose a multi-faceted approach to improve predictive accuracy and optimize glucose regulation. Our research is structured around three specific aims: Aim 1: Develop biologically feasible ML models for glucose prediction. We will enhance predictive accuracy by integrating multi-fidelity and multi-modal data, including signals from insulin pumps, fitness trackers, and continuous glucose monitors. Aim 2: Construct mathematically consistent computational models for glucose-insulin dynamics. We will develop hybrid models that combine data-driven ML approaches with physiological models based on ordinary differential equations (ODEs) to improve precision and robustness. Aim 3: Design robust and optimized glucose control algorithms using offline reinforcement learning (RL). By leveraging offline RL, transfer learning, and ensembling techniques, we aim to create a cost-effective and reliable control system for patient-specific glucose management. Successful completion of this project will significantly enhance our ability to predict and regulate glucose levels in T1D, leading to improved patient outcomes. Our approach will integrate real-world, multi-source data to create accurate digital twins of glucose-insulin dynamics, enabling personalized interventions. These innovations will contribute to the next generation of AI-driven diabetes management, ultimately improving therapeutic strategies and clinical decision-making. |  |