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 third annual Computational Medicine Pilot Grant Awards.

The winning teams include:

“Statistical Inference for Hi-C data – Is It Really There?”

  • Co-PI: J.S. Marron, Ph.D., Amos Hawley Distinguished Professor, Department of Statistics and Operations Research
  • Co-PI: Hyejung Won, Ph.D., Assistant Professor, Department of Genetics

Hi-C is a genomic technology for understanding 3-d structure of chromosomes within the nucleus. A wide range of analytic tools have been developed to detect well-established features of chromosome conformation, such as loops, topological associated domains (TADs), and compartments. However, each tool was specialized to detect a given specific structure, and no tool was available to detect multiple structures using a single algorithm. We aim to revolutionize the analysis of Hi-C data by developing a tool set for integrating the discovery of features by visualization with statistical inference for determining which represent strong underlying signal and which are sampling artifacts. That inference will be inspired by the earlier methodology of Significance in Scale Space. Funding (student support and a computer) is requested to generate compelling preliminary results for writing an NIH R01 proposal.

“A dual screening approach for engineering durable and effective T lymphocyte responses”

  • Co-PI: Natalie Stanley, Ph.D., Assistant Professor, Computer Science
  • Co-Pi: Justin Milner, Ph.D., Assitant Professor, Microbiology and Immunology

A hallmark of cancers, chronic infections, and autoimmunity is a steady decline in the protective potential of CD8 T-cell responses, known as T-cell exhaustion. Bioinformatics strategies can be applied to characterize the genetic programs underlying the diverse ontogeny of CD8 T-cell states, but existing methods are limited in that they 1) fail to explicitly account for temporal information and 2) do not quantitatively characterize how fate committed cells are toward a particular terminal state. To develop rigorous bioinformatics strategies to categorize the fate-commitment of individual cells, we will use single-cell RNA sequencing data collected at 11 post Lymphocytic Choriomeningitis Virus (LCMV) infection timepoints. By combining trajectory inference and machine learning approaches, we will identify gene expression patterns at early post-infection timepoints predictive of terminal cell-fates. Genes driving predictions in the machine learning model will be used to design a targeted screen to evaluate the effects of knocking down genes on terminal CD8 T-cell fates. We anticipate that the developed methods will advance the way in which longitudinal single-cell data are used to identify critical precursors and their genetic programs of downstream deleterious states, such as CD8 T-cell exhaustion.

“A novel experimental/computational strategy to identify and characterize drug-resistant subpopulations in primary human tumors”

  • Co-PI: Sam Wolff, Ph.D., Assistant Professor, Department of Genetics
  • Co-PI: Philip Spanheimer, PhD., Assistant Professor, Department of Surgery

Heterogeneity within human cancers is associated with resistance to therapy, which cannot be dissected by bulk genomic methods. Single-cell transcriptomic approaches have been developed; however, these have not been coupled to experimental systems where live human tumors can be challenged with therapeutics. We have recently adapted a proteomic imaging platform (4i) to quantify single-cell differences in signaling and cell cycle regulators. This innovative approach could revolutionize how we assess emerging therapeutic resistance in human tumors; however, two key challenges remain. First, our platform must accurately recapitulate distinct cellular subpopulations within human tumors and represent in vivo growth. Second, innovative computational techniques are needed to analyze this new high dimensional dataset. Specifically, these computational techniques must: (1) capture nuanced distributions of single cells including rare subpopulations; (2) numerically integrate single-cell data across different tumor samples; and (3) identify resistant subsets of cells within a low dimensional, interpretable framework. In Aim 1 of this proposal, we apply 4i to a primary human ER+/HER2-breast tumor using novel procurement, culture, staining, and image processing protocols to demonstrate recapitulation of tumor subpopulations and proliferative markers. In Aim 2, we use our experimental model to challenge explanted ER+/HER2- breast tumors with tamoxifen and develop a novel computational strategy to assess responsive and resistant subpopulations. This pilot project will generate both new single cell proteomic data of primary human tumors as well as innovative computational methods for identifying resistant subpopulations of tumor cells, positioning our team to compete for multiple extramural funding awards including PAR-22-099 and PAR-22-131.

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