The advent of spatial transcriptomics has opened new avenues for exploring spatial heterogeneity in gene expression across tissues. However, effectively quantifying the influence of specific cell populations on their niche remains a key challenge. Here, we present stGrads (Spatial Transcriptomic Gradients), a computational framework designed for both spot-size and bin-size spatial data to characterize the spatial gradients induced by specific cell types at both gene expression and cell composition levels. By integrating spatial proximity modeling with attenuation-based signal propagation, stGrads computes multiple distance-dependent metrics, including expression gradients, compositional shifts, and interaction strengths, to reveal local patterns of cellular responses. Meanwhile, stGrads could identify the gradient-related genes for downstream work. We demonstrate the utility of stGrads on multiple spatial transcriptomic datasets, successfully identifying disease-associated spatial gradients. stGrads offers a generalizable and efficient tool for characterizing spatial interactions and functional responses in complex tissue environments.
Fu et al. (Sun,) studied this question.