Spatial Transcriptomics

Spatial transcriptomics studies where genes are active in intact tissue, using assays that measure expression at spots or individual cells and linking molecular profiles to tissue architecture. Computational approaches include spatial graphs and transformers for domain discovery and imputation, optimal transport for aligning tissue sections, and multimodal models that combine expression with histology or other omics. These methods support cell-neighborhood analysis, developmental mapping, and prediction of molecular patterns from pathology images, with data from platforms such as Visium, Xenium, MERFISH, and Stereo-seq. A central challenge is balancing resolution and assay noise with generalization across tissues and platforms; predictive scores may also fail to capture spatial heterogeneity or usefulness for biological discovery.