DiffGSP: reversing mRNA diffusion to unlock high-fidelity spatial transcriptomics

Abstract

Spatial transcriptomics enables gene expression profiling within intact tissues while preserving spatial context, providing unprecedented insights into cellular organization and function. However, mRNA diffusion during tissue processing can cause transcripts originating from adjacent cells to be captured at a given spot. This spatial misalignment challenges a fundamental premise of spatial transcriptomics that measured gene expression faithfully corresponds to its spatial origin, thereby compromising spatial fidelity and potentially biasing biological interpretation. Here, we present DiffGSP, a physics-informed framework that integrates Fick's law with graph signal processing to explicitly model diffusion-induced distortions and recover the underlying spatial gene expression landscape. Comprehensive benchmarking across diverse datasets and evaluation metrics demonstrates the consistent ability of DiffGSP to restore spatial gene expression patterns. By computationally reversing diffusion-induced distortions, DiffGSP enables the discovery of fine anatomical structures in the mouse brain, spatially organized gene modules in the kidney, intratumoral heterogeneity in colorectal cancer, and tertiary lymphoid structures in lung adenocarcinoma. Built on a physically interpretable framework and broadly applicable to sequencing-based spatial transcriptomics technologies, DiffGSP improves the fidelity of spatial gene expression reconstruction and enables more reliable biological interpretation.