The paper introduces SPLISOSM, a novel computational framework designed to identify spatial isoform variability across diverse spatial transcriptomics platforms. By employing spectral graph theory and kernel-based statistical tests, the method maintains high sensitivity while effectively filtering out noisy signals in large-scale biological datasets. A significant innovation includes a conditional independence framework that prevents false positives caused by spatial confounding, allowing researchers to distinguish true biological patterns from correlated noise. The authors demonstrate the tool's utility by uncovering cell-type-specific splicing and evolutionarily conserved gene regulation in both mouse and human brain tissues. Furthermore, the analysis reveals how the tumor microenvironment drives distinct transcript usage in glioblastoma, particularly regarding immune response and metabolic pathways. Ultimately, this work provides a robust mathematical foundation for exploring the spatial architecture of RNA processing in complex tissues.

References:

• Su J, Qu Y, Schertzer M, et al. Mapping isoforms and regulatory mechanisms from spatial transcriptomics data with SPLISOSM[J]. Nature Biotechnology, 2026: 1-12.