Machine learning methods to identify tumor-immune microenvironments in spatial transcriptomics tumor data
Barbara Engelhardt, PhD
Professor (Research), Department of Biomedical Data Science and, by courtesy, of Statistics
Friday, May 17, 2024
11:00am - 12:00pm
James H. Clark Center, Room S360, 3rd floor next to the Coffee Shop
Zoom link
Abstract:
Tumor-immune microenvironments (TIMEs) capture the state of cancer and the body's response to a solid tumor, and characterizing these TIMEs can lead to better treatment decision-making for the patient. Our work describes a machine learning method for identifying and characterizing TIMEs across spatial transcriptomic data, or 2D slices of solid tumors with spatially-resolved transcript markers identified at single-cell resolution. We apply our approach to three Merfish samples of colorectal tumors, and we find known and unknown TIMEs, and we characterize different types of TIMES and highlight their spatial organization and function.