What happened
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. Update (2026-09-15): DiffGSP: reversing mRNA diffusion to unlock high-fidelity spatial transcriptomics . 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 Update (2026-09-22): Selective depletion of upper-layer somatostatin interneuron subtypes in schizophrenia . Schizophrenia (SCZ) is associated with cortical GABAergic dysfunction, but whether inhibitory interneurons are lost or persist in an altered molecular state remains unresolved. Here, we harmonized seven prefrontal post-mortem single-nucleus RNA-seq datasets onto a fine-grained taxonomy of cortical cell types and meta-analyzed their gene expression and abundance changes in SCZ (298 controls, 171 SCZ). First, we find a subclass-wide reduction of SST mRNA within somatostatin (Sst) neurons. Second, we find reduced abundance (depletion) of a subset of upper-layer Sst interneurons and increased abundance of L6b excitatory neurons, with both changes confirmed in spatial transcriptomics (12 controls, 12 SCZ). Notably, SCZ genetic risk is enriched in the most depleted Sst cells. Depleted Sst subtypes highly express HCN1, partially correspond to primate-specialized CALB1-expressing double-bouquet cells, and are among the cells lost earliest in Alzheimer's disease. These upper-layer Sst interneurons constitute an intrinsically vulnerable population and a promising target for neur Update (2026-09-22): Spatial transcriptomics reveals selective vulnerability of cardiac neural crest-derived glial cells in human myocardial infarction . Cardiac neural crest-derived glial cells (CNGs) are recently identified glial cells that support cardiac autonomic innervation and maintain sympathetic-parasympathetic balance. Their fate in myocardial infarction (MI) is unknown. Here, we analyzed 16 human cardiac spatial transcriptomes spanning normal myocardium (n=4), infarct zone (n=5), border zone (inner, n=1; outer, n=3), and remote zone (n=3). CNGs (S100B+/GFAP+, CD68- spots) were selectively lost along a spatial gradient from healthy tissue to the infarct core (Spearman rho = -0.92, p = 4.5 x 10-7). Tissue-area-corrected analysis confirmed that CNG loss (26% retained in infarct zone vs. normal; p = 0.016) vastly exceeded overall tissue loss, indicating selective vulnerability rather than bulk tissue destruction. Mechanistically, oxidative phosphorylation (OXPHOS) activity in CNG-positive spots correlated positively with CNG survival (partial rho = 0.65, p = 0.009 after controlling for spot composition), consistent with a high metabolic demand underlying their ischemic