In the last decade, single-cell RNA sequencing (scRNA-seq) transformed biology by moving from "bulk" averages—where thousands of cells are blended into a single molecular "smoothie"—to the resolution of individual cells. However, this progress came at a cost: to sequence a cell, you must first rip it out of its home. This dissociation destroys the spatial context, the "neighborhood" that defines how a cell behaves, who it talks to, and how it contributes to the larger organ.
Spatially resolved transcriptomics (SRT) fixes this by mapping the "what" (gene expression) directly onto the "where" (physical coordinates). It allows researchers to build what Muzlifah Haniffa and Deanne Taylor describe in A roadmap for the Human Developmental Cell Atlas as a "Google Maps" of the body—a high-resolution coordinate system where molecular activity is anchored to anatomical structure.
The Neighborhood is the Message
The core intuition behind SRT is that a cell’s identity is not just a product of its lineage, but a dynamic response to its environment. In the liver, for instance, hepatocytes perform different tasks depending on their proximity to blood vessels, a phenomenon known as zonation. Franziska Hildebrandt and Sami Saarenpää demonstrated in Spatial Transcriptomics to define transcriptional patterns of zonation that by preserving spatial coordinates, we can see these functional "divisions of labor" that are invisible in dissociated data.
This environmental influence is even more stark in disease. In pancreatic cancer, Natalie Landon‐Brace and colleagues showed that low oxygen levels (hypoxia) in specific niches can force cancer cells to transition into more aggressive subtypes (Analysis of an engineered organoid model of pancreatic cancer). Without spatial data, we might assume these aggressive cells are a separate genetic "breed"; with SRT, we see they are simply cells reacting to a local lack of air.
How to Index a Tissue
To map transcripts to space, researchers use two primary strategies: physical barcoding and computational reconstruction.
1. Physical Barcoding: The Molecular GPS Techniques like Stereo-seq and Visium use glass slides or silicon chips covered in millions of microscopic "capture spots." Each spot contains a unique DNA barcode that acts like a GPS coordinate. When a tissue slice is placed on the slide, the mRNA from the cells migrates down and binds to these barcodes.
Ao Chen and Yidi Sun used this approach in Global Spatial Transcriptome of Macaque Brain at Single-Cell Resolution to create a 3D blueprint of the primate brain. A particularly clever iteration is HiFi-Slides, developed by Ekko Zhu, which "upcycles" used Illumina sequencing flow cells. By repurposing the high-density DNA clusters already present on the flow cell, researchers can achieve sub-micron resolution—fine enough to see where genes are expressed inside a single cell (High-Resolution Characterization of Human Brain Cortex).
2. Computational Reconstruction: The Spatial Fingerprint Perhaps the most surprising discovery in this field is that space is often "hidden" inside the gene expression data itself. Anna C. Schaar and Alejandro Tejada-Lapuerta introduced Nicheformer, a foundation model trained on over 100 million cells. They found that a cell’s transcriptome contains a "spatial fingerprint"—a signature of its original neighborhood density and composition—allowing the model to predict where a cell came from even after it has been dissociated (Nicheformer: A Foundation Model for Single-Cell and Spatial Omics).
Decoding the Geography of Disease
The ultimate goal of SRT is to understand how the "geography" of a tissue determines the outcome of a disease. In oncology, this means moving beyond counting immune cells to mapping "immune topographies."
For example, Valeria Cancila and Giorgio Bertolazzi used high-resolution SRT to show that aggressive B-cell lymphomas create "immune-silent" zones that physically exclude T-cells, preventing the body from attacking the tumor (Aggressive B cell lymphomas retain ATR-dependent determinants of T cell exclusion). Similarly, Emelie Berglund and Jonas Maaskola discovered that the molecular boundaries of prostate cancer often extend far beyond what a pathologist can see under a microscope, revealing a "division of labor" between the tumor's metabolic core and its inflammatory edge (Spatial maps of prostate cancer transcriptomes).
By integrating these spatial maps with deep learning tools like GraphST (Yahui Long et al., Spatially informed clustering) or HyperSTAR (Yi Liao et al., Unveiling Tissue Structure), researchers are now able to denoise complex data and identify the subtle "handshakes" between cells that drive cancer progression, neuroinflammation in Alzheimer's, and the success or failure of organ transplants.