Decoding Cancer Heterogeneity Through Multi-Modal Biomedical Data

 Yuanning Zheng, PhD
  Postdoctoral Scholar at the Center for Biomedical Informatics Research (BMIR), School of Medicine, Stanford University

Friday, April 18, 2025
11:00am - 12:00pm  
James H. Clark Center, Room S360, 3rd floor next to the Coffee Shop
Zoom Link
Flyer

Abstract
Cancer is a highly heterogenous disease. Despite recent advancements in therapeutic strategies, clinical responses vary significantly among patients. In non-small-cell lung cancer, for instance, the objective response rate to immune checkpoint inhibitors ranges from 15% to 45%, indicating that only a subset of patients benefits from these treatments. Identifying reliable biomarkers to predict treatment response can optimize healthcare resources, reduce unnecessary exposures to side effects, and improve patients' quality of life. Current available methods have enabled us to measure cancer heterogeneity through multi-modal and multi-omics biomedical data. However, harnessing the complementary strengths of these datasets still represents a challenge due to their high dimensionality and the presence of missing data modalities. To tackle these challenges, our research focuses on developing bioinformatics tools and AI models to identify clinically relevant disease phenotypes. Key contributions include: (1) SEQUOIA (Nature Communications, 2024), a vision transformer model that predicts transcriptomic profiles from H&E-stained whole-slide images across 16 cancer types; (2) GBM360 (Nature Communications, 2023), a deep-learning framework that integrates spatial transcriptomics and H&E-stained histology images to elucidate spatial cellular architectures associated with glioblastoma prognosis; and (3) EpiMix (Cell Reports Methods, 2023), a bioinformatics tool for population-level analysis of DNA methylation and gene expression data. This talk will further cover our insights derived from spatial molecular technologies, including multiplex imaging (CODEX), Visium and CosMx. By driving computational advancements, we harness the power from (epi)genomics, transcriptomics and medical imaging data to decode tumor heterogeneity at molecular, tissue and patient levels.

Biography
Dr. Yuanning Zheng is a postdoctoral scholar at the Center for Biomedical Informatics Research (BMIR) at Stanford University School of Medicine. His research centers on developing computational methods that integrate (epi)genomics, transcriptomics, and medical imaging data to advance our understanding of cancer heterogeneity. During his postdoctoral training with Dr. Olivier Gevaert, he developed novel computational approaches to elucidate the spatial cellular organization and signaling interactions within tumor microenvironments that affect therapeutic response in cancer patients. Dr.Zheng obtained his Ph.D. from Texas A&M University in 2021, where his research focused on investigating the epigenetic and cellular mechanisms through which early-life environmental exposures influence mammary cancer risk.