Why Location Matters
Single-cell RNA sequencing gave us the ability to profile gene expression in individual cells. But in doing so, it destroys the very thing that makes tissues work: spatial organization. When you dissociate a tumor into a suspension of single cells, you lose all information about which cells were neighbors, which were near blood vessels, and which were at the invasive margin. You lose the microenvironment.
Spatial transcriptomics restores this lost context. By measuring gene expression while preserving the physical location of each measurement within a tissue section, spatial methods let us study biology as it actually exists — in organized, interacting communities of cells within complex tissue architectures.
The Technology Landscape
Spatial transcriptomics technologies can be broadly divided into two categories: sequencing-based and imaging-based approaches.
Sequencing-Based Methods
- 10x Visium: The most widely adopted platform. Uses spatially barcoded oligonucleotide arrays to capture mRNA from tissue sections at ~55 micrometer spot resolution (roughly 1-10 cells per spot). Genome-wide coverage but not single-cell resolution
- 10x Visium HD: The next-generation platform offering 2-micrometer resolution — effectively single-cell — while maintaining whole-transcriptome coverage. A major step forward for the field
- Slide-seq / Slide-seqV2: Uses DNA-barcoded beads packed into a dense array, achieving ~10 micrometer resolution with whole-transcriptome coverage
- Stereo-seq: Developed by BGI, uses DNA nanoball technology to achieve subcellular resolution spatial transcriptomics at scale
Imaging-Based Methods
- MERFISH (Vizgen): Multiplexed error-robust fluorescence in situ hybridization. Images hundreds to thousands of genes at single-molecule, subcellular resolution. The gold standard for spatial resolution, but limited to a predefined gene panel
- seqFISH+: Similar to MERFISH, uses sequential rounds of hybridization and imaging to profile thousands of genes at subcellular resolution
- CosMx (NanoString): Single-molecule imaging platform that profiles up to 6,000 genes per cell in intact tissue with subcellular resolution
- Xenium (10x Genomics): In-situ sequencing platform targeting hundreds of genes at single-cell resolution with automated, high-throughput workflows
The Resolution-Coverage Tradeoff
There's a fundamental tradeoff between spatial resolution and transcriptome coverage. Sequencing-based methods like Visium cover the whole transcriptome but at lower resolution. Imaging-based methods like MERFISH achieve subcellular resolution but are limited to predefined gene panels. Visium HD and Stereo-seq are beginning to bridge this gap, offering both high resolution and broad coverage.
Computational Challenges
Cell Segmentation
For imaging-based methods, the first challenge is identifying individual cells in the tissue image. This requires sophisticated cell segmentation algorithms that can handle the complexity of real tissue — overlapping cells, irregular shapes, varying cell sizes, and dense packing. Deep learning approaches (Cellpose, StarDist, Baysor) have made significant progress, but segmentation errors still propagate through all downstream analyses.
Spot Deconvolution
For Visium and similar methods where each spot contains multiple cells, computational deconvolution is needed to estimate the cell type composition of each spot. Tools like cell2location, RCTD, and stereoscope use reference single-cell datasets to infer the mixture of cell types present at each spatial location. The accuracy depends heavily on the quality and completeness of the reference dataset.
Spatial Domain Identification
Beyond individual cell types, tissues are organized into functional domains — tumor core vs. margin, germinal center zones, cortical layers. Identifying these spatial domains requires algorithms that consider both gene expression and spatial proximity. Methods like BayesSpace, SpaGCN, and STAGATE use spatially-aware graph neural networks to identify tissue domains that correspond to biologically meaningful regions.
Cell-Cell Communication
One of the most exciting applications of spatial data is mapping cell-cell communication in situ. Tools like COMMOT, SpaTalk, and MISTy can infer ligand-receptor interactions between spatially proximate cells, revealing the signaling networks that govern tissue behavior. This is particularly valuable in tumor immunology, where the spatial arrangement of immune and tumor cells determines treatment response.
Data Scale and Storage
Spatial transcriptomics generates enormous datasets. A single MERFISH experiment can produce terabytes of imaging data. Visium HD datasets contain millions of spatial measurements. Efficient data structures, out-of-core computing, and scalable infrastructure are essential. The AnnData format (with its SpatialData extension) and Zarr-based storage are becoming the standards for managing spatial datasets.
Applications Transforming Biology
Tumor Microenvironment Mapping
Spatial transcriptomics is revealing the complex cellular ecosystems of tumors. Researchers can now map the spatial organization of tumor cells, immune cells, stromal cells, and vasculature, and correlate these patterns with treatment response and patient outcomes. This spatial context is proving essential for understanding why immunotherapy works in some patients but not others.
Neuroscience
The brain is perhaps the most spatially organized organ, with distinct cell types arranged in precise layers and regions. Spatial transcriptomics is creating the most detailed molecular atlases of the brain ever produced, revealing new cell types and spatial gene expression patterns associated with neurological diseases.
Developmental Biology
Understanding how tissues and organs develop requires knowing not just what genes are expressed, but where. Spatial transcriptomics applied to developing embryos is revealing the molecular programs that guide cell fate decisions in their native spatial context.
Building a Spatial Analysis Platform
At Next Generation Consulting, we've built spatial transcriptomics analysis platforms for pharmaceutical clients studying tumor microenvironments, neuroscience teams mapping brain regions, and developmental biologists studying organogenesis. Our platform architecture typically includes:
- Data ingestion: Automated pipelines for processing raw data from Visium, MERFISH, Xenium, and CosMx platforms
- Quality control: Spatial-aware QC metrics including tissue detection, spot quality, and per-gene spatial autocorrelation
- Analysis workflows: Modular Nextflow pipelines for normalization, clustering, deconvolution, domain identification, and cell communication analysis
- Visualization: Interactive web applications for exploring spatial gene expression patterns overlaid on tissue histology
- Integration: Methods for combining spatial data with matched single-cell RNA-seq for enhanced resolution and annotation
Spatial transcriptomics is arguably the most exciting frontier in genomics today. The technology is maturing rapidly, the computational tools are becoming more powerful, and the biological insights are truly transformative. The organizations building spatial analysis capabilities now will lead the next generation of biological discovery.