Skip to main content

The UNC Computational Medicine Program, the SOM Office of Research, and the Lineberger Comprehensive Cancer Center announce the winners of the fourth annual Computational Medicine Pilot Grant Awards.

The winning teams include:

“Using Convolutional Neural Networks to Detect Extrachromosomal DNA and Track its Dynamics during Drug Treatment”

  • Co-PI: Elizabeth Brunk, Ph.D., Assistant Professor, Department of Pharmacology and Department of Chemistry
  • Co-PI: Ashok Krishnamurthy, Ph.D., Lead Faculty, Master of Applied Data Science and Director, Renaissance Computing Institue

Extrachromosomal DNA (ecDNA) are large (~kilo to megabase) acentric, atelomeric, circular DNAs that are established cytogenetic markers for malignancy, hard-to-treat tumors, and drug resistance. It has been proposed that cells achieve drug resistance by modulating ecDNA in numerous ways: by increasing ecDNA counts, eliminating ecDNAs, or reintegrating ecDNAs in chromosomes within homogenous staining regions (HSRs) and restoring them upon drug withdrawal. However, what remains unclear is how cell-to-cell variation in ecDNA contributes to response plasticity. One of the most unique properties of ecDNA is its ability to generate extreme cell-to-cell variation in copy number of key oncogenes, which may range in zero to hundreds to thousands of copies per cell. Such heterogeneity is not physically possible to achieve with HSRs, therefore, ecDNA may confer unique advantages by rapidly increasing the genetic diversity of a cell population. Understanding how this diversity contributes to drug resistance is an urgent unmet need and would change the ways we design drugs that counteract drug resistance. This project will develop computational algorithms to quantify changes in ecDNA and HSRs during drug treatment through a collaboration between the Brunk lab, RENCI, and an undergraduate-led machine learning club, AI@UNC. Together, we will develop a convolutional neural network approach to fast-track image processing and ecDNA counting in FISH images. This research integrates computational approaches, such as computer vision and neural networks, to enhance biomedical cancer research and pharmacology by establishing automated methods for detecting highly aberrant oncogene amplifications in cancer cells. Development of this algorithm will enable our team to process and analyze our cytogenetics data, and compete for our first extramural grant. In the longer term, these efforts will enable the development of a high-throughput platform to systematically assess the role of ecDNA plasticity in the development of drug resistance.

“Elucidating Control Mechanism of Tau Propagation for Alzheimer’s Disease”

  • Co-PI: Guorong Wu, Ph.D., Associate Professor, Department of Psychiatry
  • Co-Pi: Todd Cohen, Ph.D., Associate Professor, Department of Neurology

Mounting evidence shows that Alzheimer’s disease (AD) is characterized by the propagation of tau aggregates throughout the brain in a prion-like manner. Since current pathology imaging technologies only provide a spatial brain mapping (snapshot) of tau accumulation, computational modeling becomes indispensable in analyzing the spatiotemporal propagation patterns of widespread tau aggregates. Recently, we have developed a cutting-edge computational approach to uncover the spreading pathway of tau propagation from longitudinal neuroimages. In this project, we aim to integrate the insight of control theory and the power of machine learning to understand the control mechanism of tau propagation in the context of amyloid-beta (Aβ) accumulation and neuroinflammation burden. Specifically, we will develop a physics-informed deep model for the COntrol MEchanism of Tau Spreading (coined COMETS). In scientific exploration, we will first validate our deep model on PS19 Transgenic mice (the gold standard in the field). Then, we will apply COMETS to longitudinal PET images of human brain from public datasets. By doing so, we will investigate (i) the global and regional controllability of how Aβ-tau dynamics determine spreading patterns of tau aggregates, and (ii) the mechanistic role of neuroinflammation in both animal and human studies. Expected outcomes: A system-level brain mapping of interactions between AD biomarkers that underlines diverse trajectories of cognitive decline in the clinic, which sets the stage for large-scale NIH R01 grants.

 

Thank you to all of the applicants and congratulations to the winners.