RET Projects

We are excited to host the Summer RET Site under the NSF Grant titled “RET Site: Teachers As Researchers in Computing Classrooms (TARCC) Bridging Gaps in High School Computing Education in Western Wisconsin and Eastern Minnesota” in 2025.

The RET site will support research for 10 teachers from Western Wisconsin and Eastern Minnesota each summer engaging them in a collaborative research experience with the central theme of designing computationally intensive algorithms and simulation models as well as reflecting on their intensive research experiences through evidence-based teaching approaches and the scholarship of teaching and learning (SoTL). The research will be conducted using UW-Eau Claire's Blugold High-Performance Computing facility and on campuses of UW-Eau Claire (UWEC), UW-Stout (UWST) and UW-River Falls (UWRF).

Computer Science Projects at UW-River Falls

Deploying Scalable Bioinformatics using Cloud Computing

At the University of Wisconsin-River Falls (UWRF), several research labs utilize bioinformatics to analyze gene expression in plant species such as Rose and Ninebark. RNA sequencing (RNAseq) generates vast amounts of data, making cloud computing a key tool for processing, analyzing, and visualizing these large datasets. This project employs cloud-based solutions like ElasticBLAST, offered by platforms such as Amazon Web Services (AWS) and Microsoft Azure, to perform functional genomics analysis and RNAseq contamination checks.

By leveraging cloud infrastructure, this research enables the detection of contaminants in sequencing data and facilitates the storage and processing of data from third-generation sequencing platforms like Oxford Nanopore. Faculty members, including Dr. Varghese, who have extensive experience in cloud-based computational techniques, guide participants through the development of bioinformatics modules and Jupyter notebooks. This program offers educators and students an opportunity to gain hands-on experience in applying cloud computing to solve real-world bioinformatics problems.

Markov State Models of Ion Channels

Ion channels are essential proteins found in nearly every cell, responsible for regulating ion movement and generating electrical activity, such as action potentials in nerve and heart cells. At the University of Wisconsin-River Falls (UWRF), Dr. Varghese leads research focused on modeling the human Ether-a-gogo Related Gene (hERG) ion channel, using experimental data from collaborators in Sydney, Australia. The project aims to develop Markov state models that fit this extensive dataset.

Participants in this project will gain hands-on experience with mathematical modeling and regression analysis techniques. They will also explore optimization methods like particle swarm optimization to improve model fitting. This research provides a unique opportunity to work with cutting-edge techniques in computational biology and biophysics.

Simulation of Electrical Activity in Cardiac Cells

At the University of Wisconsin-River Falls (UWRF), Dr. Varghese leads research on simulating the electrical activity in heart cells using numerical methods. Each cell’s behavior is modeled by a complex system of approximately 30 nonlinear ordinary differential equations, which are solved numerically due to the lack of analytical solutions. Participants will explore numerical methods, such as implicit ODE solvers, using Python or Java to simulate both normal and pathological cardiac physiology.

In addition to conducting simulations, participants will develop modules on time integration of differential equations and stability analysis. These educational modules will introduce high school students to key concepts in computational physiology and numerical analysis, bridging advanced topics with accessible classroom learning.

Using bioinformatics tools to analyze genomic sequence data of Rose, R. Setigera

The Rose RNAseq project, led by Dr. Varghese in collaboration with faculty from the Department of Plant and Earth Sciences at the University of Wisconsin-River Falls (UWRF), focuses on the gene expression of the dioecious species Rosa setigera, native to North America. This research involves isolating mRNA and sequencing it using the Oxford Nanopore third-generation sequencing platform, with the resulting long-read DNA sequence data deposited in GenBank (NCBI GenBank BioProject PRJNA1042965) for public accessibility.

By comparing R. setigera gene sequences with those of related rose species that have previously been sequenced, the project aims to elucidate the unique phenotypes of R. setigera. Future work will involve sequencing the entire R. setigera genome to further investigate gene transcription and expression. Participants will gain practical experience in obtaining biological sequence data, performing sequence comparisons in existing databases, and utilizing tools in Python and bioinformatics at the UWEC Blugold Center for High Performance Computing.

Computer Science Projects at UW-Eau Claire

Evaluation of DNA Methylation Markers using Machine Learning Algorithms

This project examines DNA methylation, a key process that impacts gene accessibility and expression by adding methyl groups to cytosines in DNA (Moore, 2013). Participants will develop a scalable deep-learning framework to analyze high-dimensional genomic datasets, aiming to identify methylation sites in the human genome associated with tumors. The program will focus on implementing a dynamic feature selection algorithm, utilizing techniques such as the ANOVA F-test and Random Forest to extract relevant features, followed by creating a diagnostic prediction model using a fully connected deep neural network.

The program will commence with a two-week introduction to machine learning, followed by two weeks on deep learning fundamentals. Data from the Cancer Genome will be utilized for model construction, leveraging the Blugold Center for High Performance Computing (BCHPC) at UWEC to manage the high dimensionality of datasets, which can include up to 485,000 methylation sites per sample.

Deep Learning for Diagnosis of Pancreatic Cancer to Advance Patient Healthcare

Pancreatic Ductal Adenocarcinoma (PDAC) is one of the most aggressive cancers, with a 5-year survival rate of only 6%. As the number of cases continues to rise, this project at the University of Wisconsin-Eau Claire (UWEC) explores advanced imaging and deep learning techniques to improve the diagnosis of PDAC. Participants will begin by learning essential image processing techniques, including normalization and DICOM format handling. In later stages, they will build a UNet-based Convolutional Neural Network (CNN) to analyze CT scan images from public datasets such as the Medical Segmentation Decathlon and the Cancer Imaging Archive.

The project emphasizes hands-on experience with deep learning models and the use of GPU nodes on UWEC’s BCHPC cluster. Participants will gain practical knowledge in both computational techniques and their application to improving patient outcomes in healthcare.

Using Dynamic Feature Selection (DFS) to Address Cyber Threats

As cyber threats continue to evolve, it becomes crucial to develop adaptable and efficient detection methods for malicious internet traffic. At the University of Wisconsin-Eau Claire (UWEC), this project focuses on utilizing Dynamic Feature Selection (DFS) to address the challenges of threat detection. Traditional machine learning methods for cybersecurity often struggle with highly imbalanced datasets and large feature dimensions, which can limit real-time detection capabilities. To overcome these challenges, this project applies DFS techniques to automate feature reduction, improving computational efficiency and predictive performance.

Participants will explore DFS methods, including Univariate Feature Selection, Correlated Feature Elimination, Gradient Boosting, and Wrapper Methods. These techniques will be applied to well-known cybersecurity datasets such as NSL-KDD and UNSW-NB15, forming the basis for a machine learning pipeline that enhances the detection of harmful internet traffic. Teachers will spend the first two weeks learning machine learning algorithms and the remaining time building the DFS strategy, training models, and optimizing hyperparameters.