Integrative Computational Models for Decoding Disease Mechanisms and Predicting Drug Synergies in Spatial Transcriptomics

PROJECT SUMMARY/ABSTRACT Cellular Spatial Organization (CSO)—the spatial arrangement of cells within tissues—plays a critical role in shaping the tissue microenvironment, disease progression and therapeutic outcomes, particularly in complex conditions such as cancer, neurodegenerative diseases, and inflammatory disorders. Recent advancements in Spatially Resolved Transcriptomics (SRT) offer unprecedented capabilities to map gene expression at or near single-cell resolution while preserving spatial context. However, the high costs and technical complexities associated with SRT limit its application in large-scale, phenotype-rich studies. This challenge necessitates the integration of SRT data with accessible clinical datasets, such as bulk RNA-seq from large patient cohorts, to reveal critical insights into how CSOs drive disease mechanisms. Current computational methods, however, primarily focus on unsupervised analysis of spatial patterns in SRT data, limiting our ability to link CSO patterns with essential clinical outcomes such as patient survival, therapeutic resistance, and drug combination responses. My research program will address this gap by developing scalable statistical and computational methods that integrate SRT data with clinical outcomes, enabling us to uncover how CSO influences disease progression and treatment efficacy. Over the next five years, my lab will: (1) Develop a matrix factorization-based model to integrate SRT data with bulk RNA-seq and clinical phenotypes to identify CSOs that are associated with key outcomes, including survival and therapeutic response, and provide insights into how spatial cellular architecture contributes to disease progression. (2) Design a deep learning framework that combines SRT with pharmacogenomic data and drug chemical features to predict how spatial heterogeneity affects drug responses, with the goal of identifying synergistic effects in complex microenvironments. (3) Develop a spatially informed polygenic regression model that integrates Genome-Wide Association Study (GWAS) data with SRT to identify disease-associated CSOs, revealing how genetic variants influence spatial cellular organization and highlighting novel therapeutic targets. To ensure robustness and clinical relevance, we will collaborate with leading biomedical research laboratories at Brown University, applying our computational methods to answer key biological questions, such as understanding cancer progression and drug responses. By bridging spatial genomics and clinical data, our work will address critical gaps in linking CSO to disease outcomes, enhancing precision medicine and advancing therapeutic strategies.