Spatial Transcriptomics Explained: Mapping Gene Expression in Tissue

Spatial Transcriptomics Explained: Mapping Gene Expression in Tissue

TL;DR: Spatial transcriptomics measures which genes are active in a tissue while preserving exactly where that activity happens. Think of it as adding GPS coordinates to gene expression data. Technologies like 10x Visium and MERFISH are revealing how cells organize themselves in tumors, brains, and developing organs — insights that traditional sequencing destroys when it grinds tissue into a cell soup. In 2026, spatial transcriptomics is transforming cancer diagnostics, drug development, and our understanding of how tissues actually work.

Why Location Matters in Biology

Imagine you are an oncologist studying a breast tumor. You run a standard single-cell RNA-seq analysis and discover that the tumor contains immune cells, cancer cells, and fibroblasts. Useful, but incomplete. Where are those immune cells? Are they infiltrating the tumor core, or stuck at the periphery, unable to reach the cancer cells they are meant to kill?

That distinction changes treatment decisions. A tumor with immune cells deep inside its core may respond to immunotherapy. A tumor with immune cells trapped at the edge may not. Standard sequencing cannot tell you which scenario you are looking at, because the first step of the process — dissociating the tissue into individual cells — destroys the very spatial information you need.

Spatial transcriptomics solves this. It measures gene expression without removing cells from their original positions in the tissue. The result is a map: not just which genes are active, but exactly where they are active, down to individual cells or even subcellular compartments.

The field earned recognition when Nature Methods named spatial transcriptomics its Method of the Year in 2020. Since then, the technology has matured rapidly, with multiple platforms now offering different trade-offs between resolution, gene coverage, and throughput.

How Spatial Transcriptomics Works

All spatial transcriptomics technologies face the same fundamental challenge: detecting RNA molecules while keeping track of their physical coordinates in the tissue. Two main strategies have emerged, each with distinct strengths.

Sequencing-Based Methods (Visium, Slide-seq)

10x Genomics Visium is the most widely used platform. A thin tissue slice is placed on a glass slide covered with thousands of barcoded capture spots, each 55 micrometers in diameter. RNA from the tissue binds to these spots, and the barcode on each spot encodes its position. After sequencing, you can reconstruct where each RNA molecule came from.

Visium captures the entire transcriptome — all ~20,000 genes — at every spot. The trade-off is resolution. Each 55-micrometer spot covers roughly 1 to 10 cells, so you get near-cellular but not true single-cell resolution. The newer Visium HD platform reduces spot size to 2 micrometers, achieving single-cell resolution while maintaining whole-transcriptome coverage.

Slide-seq uses a similar concept but replaces the printed array with a dense layer of DNA-barcoded beads, achieving 10-micrometer resolution. Stereo-seq, developed by BGI, pushes this further with nanometer-scale spatial barcoding.

Imaging-Based Methods (MERFISH, seqFISH)

MERFISH (Multiplexed Error-Robust Fluorescence In Situ Hybridization) takes a completely different approach. Instead of capturing RNA on a surface, it directly visualizes RNA molecules inside intact cells using fluorescent probes. Each gene is assigned a unique pattern of fluorescent signals across multiple rounds of imaging. By photographing the tissue repeatedly and decoding the signal patterns, MERFISH identifies individual RNA molecules at their exact locations.

The result is true subcellular resolution — you can see which RNA molecules are in the nucleus versus the cytoplasm. The trade-off is gene coverage: MERFISH typically measures hundreds to a few thousand genes per experiment, not the full transcriptome.

seqFISH+ achieves similar resolution with a different encoding scheme, and can detect over 10,000 genes in a single experiment.

Sequencing vs. Imaging: How to Choose

In practice, many research groups use both: Visium for initial discovery across a broad gene set, then MERFISH to zoom in on specific genes of interest at higher resolution.

What Spatial Transcriptomics Reveals That Other Methods Cannot

Tumor Microenvironments

Cancer is not just about cancer cells. Tumors are complex ecosystems of malignant cells, immune cells, blood vessels, and supporting tissue. Spatial transcriptomics maps these ecosystems in detail.

A 2023 study in Nature used spatial transcriptomics to map the tumor microenvironment in colorectal cancer, identifying distinct spatial neighborhoods where immune cells either attacked or protected the tumor. Patients whose tumors showed immune-enriched niches near the tumor border had significantly better outcomes after immunotherapy.

This kind of spatial biomarker — not just "are immune cells present?" but "where are they relative to the cancer?" — is impossible to measure without spatial transcriptomics.

Brain Architecture

The brain is the most spatially organized organ in the body. Neurons that are millimeters apart can have completely different functions. MERFISH studies of the mouse brain have identified over 5,000 transcriptionally distinct cell populations organized into precise spatial domains. The Allen Brain Cell Atlas, released in 2023, used MERFISH to create the most detailed spatial map of the mammalian brain to date.

These maps are revealing how neurodegenerative diseases disrupt specific spatial zones. In Alzheimer's research, spatial transcriptomics has shown that gene expression changes radiate outward from amyloid plaques in concentric rings, affecting different cell types at different distances.

Developmental Biology

How does a single fertilized egg become an organism with hundreds of cell types, each in the right place? Spatial transcriptomics is answering this question by mapping gene expression during embryonic development. Stereo-seq has been used to create spatial atlases of mouse embryos at multiple developmental stages, tracking how cells differentiate and migrate to form organs.

The Analysis Pipeline

Analyzing spatial transcriptomics data requires specialized tools that handle both the expression matrix and the spatial coordinates.

Preprocessing: Similar to scRNA-seq — quality control, normalization, and dimensionality reduction. Tools like Squidpy (Python) and Seurat v5 (R) handle spatial data natively.

Spatial neighborhood analysis: Identifying which cell types tend to co-locate and which avoid each other. Squidpy computes neighborhood enrichment scores and spatial autocorrelation statistics to quantify these patterns.

Cell-cell communication: Tools like SpaTalk and cell2location infer signaling between neighboring cells based on ligand-receptor expression and spatial proximity. Unlike non-spatial methods that assume all cells can interact equally, spatial approaches only consider biologically plausible interactions between nearby cells.

Deconvolution: For platforms like Visium where each spot contains multiple cells, deconvolution algorithms estimate the cell-type composition of each spot by comparing its expression profile to single-cell reference data. Cell2location and RCTD are widely used for this step.

AI integration: Foundation models are entering the spatial analysis space. Nicheformer, trained on over 110 million cells including 53 million spatially resolved profiles, can predict spatial context from dissociated single-cell data and transfer spatial annotations across tissues. This bridges the gap between the large volume of existing scRNA-seq data and the newer spatial datasets.

Spatial Transcriptomics and AI: A Natural Partnership

Spatial transcriptomics generates enormous, complex datasets. A single Visium HD experiment produces millions of data points with both expression and coordinate information. AI models are essential for extracting biological meaning from this complexity.

Graph neural networks are particularly well-suited to spatial data. They model tissue as a network where cells are nodes and spatial proximity defines edges, naturally capturing the neighborhood relationships that drive biology. Models trained on these spatial graphs can predict cell types, identify tissue regions, and discover spatial gene expression patterns that are invisible to standard clustering.

The integration with AI in genomics runs deeper. Combining spatial transcriptomics with protein structure prediction allows researchers to map not just where a gene is expressed, but where its protein product is likely active and what it interacts with in that specific tissue location.

Clinical Applications in 2026

Cancer diagnostics: Spatial transcriptomics is moving from research into clinical pathology. Companies like 10x Genomics and Vizgen are developing clinical-grade spatial profiling assays. The goal is to replace subjective visual assessment of tissue slides with quantitative spatial biomarkers that predict treatment response.

Drug development: Pharmaceutical companies use spatial transcriptomics to understand how drugs affect different regions of a tissue. A drug that works in the tumor core but fails to penetrate the tumor periphery will behave differently from one with uniform distribution. Spatial profiling reveals these patterns during preclinical testing.

Transplant medicine: Spatial transcriptomics of kidney and liver biopsies can identify early signs of organ rejection by mapping immune cell infiltration patterns, potentially detecting rejection before clinical symptoms appear.

Key Terms

Spatial transcriptomics: A family of technologies that measure gene expression while preserving the physical location of each measurement within the tissue. Combines molecular profiling with spatial mapping.

Visium: A sequencing-based spatial transcriptomics platform by 10x Genomics that captures whole-transcriptome data at near-cellular resolution using barcoded capture spots on a glass slide.

MERFISH: An imaging-based spatial transcriptomics method that identifies individual RNA molecules inside intact cells using combinatorial fluorescent labeling across multiple imaging rounds.

Deconvolution: A computational method that estimates the cell-type composition of multi-cell spatial spots by comparing their expression profiles against single-cell reference datasets.

Spatial gene expression: The pattern of where a gene is active across a tissue section. Spatial gene expression maps reveal tissue architecture and cell organization that bulk methods cannot capture.

Limitations and Challenges

Cost and throughput. Spatial transcriptomics remains expensive. A single Visium experiment costs $1,500 to $4,000, and most studies require multiple sections and replicates. This limits routine clinical use.

Tissue quality. Spatial methods require fresh-frozen or specially preserved tissue. Standard formalin-fixed paraffin-embedded (FFPE) samples, which make up the vast majority of clinical biobanks, have degraded RNA that reduces data quality. Newer protocols are improving FFPE compatibility, but it remains a limitation.

3D reconstruction. Tissues are three-dimensional, but most spatial transcriptomics methods analyze thin 2D sections. Reconstructing 3D spatial patterns from serial sections is computationally challenging and an active area of development.

Standardization. Unlike scRNA-seq, which has converged on relatively standard workflows, the spatial field is still fragmented across multiple platforms with different data formats, resolutions, and analysis tools. Efforts like the SpatialData framework aim to unify these into a common data model.

What This Means for Your DNA

Spatial transcriptomics is a research technology, not a consumer product. But its discoveries directly improve the interpretation of personal genetic data.

When you carry a genetic variant in a gene, the health implications depend heavily on where that gene is expressed. A variant in a gene active throughout the body has different consequences than one in a gene expressed only in a narrow strip of kidney tissue. Spatial transcriptomics is building the maps that connect genetic variants to their tissue-specific effects.

DeepDNA integrates insights from spatial genomics research into its DNA analysis. As spatial atlases grow more complete, the connection between a variant in your genome and its effect on specific cells in specific tissues becomes clearer and more actionable.

Understanding where genes act in tissues transforms how we interpret what genetic variants mean for health. DeepDNA applies insights from spatial genomics research to deliver more precise, tissue-aware DNA analysis — connecting your variants to the specific cell populations where they matter most. Explore your genome with DeepDNA.

This article was created with AI assistance and reviewed by the DeepDNA editorial team.