RIPPLE: replicate-aware detection of cell-type-anchored proximity gradients in spatial transcriptomics

Abstract

Cell-cell signaling shapes tissue structure and function, yet systematically decoding these circuits with spatial transcriptomics remains an open challenge. We present RIPPLE (Replicate-Aware Inference of Paracrine Profiles via Likelihood Estimation), an R package that takes a query cell type and scans all other cell types for genes whose expression varies with distance to it. On a murine lymph node 10x Xenium dataset, RIPPLE recovers the canonical T cell zone CCL21 response program in T cells and dendritic cell subsets with unanimous sign consistency across all samples. On the public CosMx non-small cell lung cancer cohort, it identifies 515 tumor-proximity gradient genes across 17 cell types, flagging a known cancer-associated fibroblast marker (IGFBP5) as the top fibroblast hit. Overall, RIPPLE delivers ranked, cell-type-resolved paracrine candidates for experimental follow-up.