Deciphering context-specific gene programs from single-cell and spatial transcriptomics data with DeCEP

Functional gene programs play a wide range of roles in health and disease by orchestrating transcriptional coregulation to govern cell identity. Understanding these intricate gene programs is essential for unraveling the complexities of biological systems; however, deciphering them remains a significant challenge. Recent advancements in single-cell RNA sequencing (scRNA-seq) and spatial transcriptomics (ST) technologies have empowered the comprehensive characterization of gene programs at both single-cell and spatial resolutions. Here, we present DeCEP, a computational framework designed to characterize context-specific gene programs using scRNA-seq and ST data. DeCEP leverages functional gene lists and directed graphs to construct functional networks underlying distinct cellular or spatial contexts. It then identifies context-dependent hub genes associated with specific gene programs based on network topology and assigns gene program activity to individual cells or spatial locations. Through evaluation on both simulated and real biological data sets, DeCEP demonstrates complementary strengths over existing methods by enabling more fine-grained characterization of gene programs within specific contexts, particularly those characterized by pronounced transcriptional heterogeneity. Furthermore, we showcase the ability of DeCEP in elucidating biological insights through case studies on normal liver tissue, Alzheimer's disease, and cancer.