Mapping Cell-to-cell Interactions from Spatially Resolved Transcriptomics Data

Cell-cell communication (CCC) is essential to how life forms, develops and functions. However, accurate, high-throughput mapping of how expression of all genes in one cell affects expression of all genes in another cell has been bottlenecked by under-developed experimental techniques and inadequate analytical designs. Here, we introduce a Bayesian multi-instance learning framework, spacia, to detect CCC from emerging spatially resolved transcriptomics (SRT) data by uniquely exploiting their spatial modality. We highlight spacia’s power to overcome fundamental limitations of popular single-cell RNA sequencing-based tools for inference of CCC, which lose single-cell resolution of CCCs and suffer from high false positive rates. Spacia unveiled how various types of cells in the tumor microenvironment differentially contribute to Epithelial-Mesenchymal Transition and lineage plasticity in tumor cells in a prostate cancer MERSCOPE dataset. We deployed spacia in a set of pan-cancer MERSCOPE datasets and derived a signature for measuring the impact of PDL1 on receiving cells from PDL1-positive sending cells. We demonstrated that this signature is associated with patient survival and response to immune checkpoint inhibitor treatments in 3,354 patients. Overall, spacia represents a notable step in advancing quantitative theories of cellular communications.