Spatial Biology: A Complete Guide to Methods, Technologies, and Applications

Spatial biology answers a question that bulk and single-cell methods cannot: not just which cells are present in a tissue, but where they sit and which neighbors they are signaling to. This guide maps the field for researchers, covering the major spatial transcriptomics, proteomics, and multiomics methods, the platforms built on them, how the resulting data gets analyzed, and where the approach is already reshaping research.

What spatial biology is and why it matters

Spatial biology is any approach that measures the molecular content of a tissue, whether that content is RNA, protein, chromatin state, or a combination of layers, while preserving where each measurement came from. For decades, that meant fluorescence in situ hybridization or immunohistochemistry, methods that could distinguish only one or two targets at a time within a tissue section.

Single-cell RNA sequencing (scRNA-seq) changed what researchers could ask about cell identity, cataloging every cell type present in a sample in unprecedented detail. But scRNA-seq requires dissociating tissue into a single-cell suspension first, a step that discards single-cell analysis in context: a cell's neighbors, its position within a niche, and the signals passing between it and the cells around it.

Spatial biology closes that gap. By combining single-cell-level molecular profiling with imaging or spatially barcoded capture, it lets researchers ask not only which molecules are present but also where they sit relative to tissue architecture and disease state. That combination is why the field has moved, in roughly a decade, from a handful of academic labs to a standard biomedical research tool available at most research institutions.

The term itself is broader than any single technology. Spatial biology encompasses any platform that ties a molecular measurement to a physical coordinate, which is why methods as different as barcoded RNA sequencing arrays, iterative antibody imaging, and label-free mass spectrometry imaging all fall under the same umbrella. What unites them is not the chemistry but the outcome: a map, rather than a list, of what a tissue contains.

That distinction matters because tissue is not a bag of independent cells. A cell's behavior is shaped by its neighbors, the gradients of signaling molecules around it, and its position within a larger structure such as a tumor nest, a lymphoid follicle, or a cortical layer. Dissociation-based methods average or discard that context entirely, which is precisely the information spatial biology was built to recover.

The rest of this guide works through that landscape: the three molecular layers spatial methods capture, the platforms built for each one, the computational steps that turn raw data into biology, where the approach is already applied, and how to weigh those options against a specific research question.

Spatial transcriptomics vs proteomics vs multiomics

The field splits into three broad molecular layers, and most spatial biology projects start by deciding which one, or which combination, actually answers the research question. Spatial transcriptomics measures RNA in place, spatial proteomics measures protein in place, and spatial multiomics attempts to capture two or more layers, sometimes from the same tissue section.

Choosing a spatial transcriptomics platform typically comes down to a split between sequencing-based methods, which capture the whole transcriptome at the cost of resolution, and imaging-based methods, which reach subcellular precision but only across a preselected gene panel. Spatial transcriptomics platform selection, resolution trade-offs, and how each method actually works are covered in more depth elsewhere in this cluster.

Spatial proteomics is a separate discipline built on antibody-based detection rather than nucleic acid probes. Multiplexed tissue imaging methods can now resolve dozens of proteins per section using fluorescence-based cycling, metal-tag mass detection, or ion-beam imaging, each with different trade-offs in plex, throughput, and tissue compatibility. Recent literature has singled out spatial proteomics advances as among the field's most consequential, citing its role in resolving tissue organization that transcriptomics alone cannot capture.

Spatial multiomics is the newest and most technically demanding of the three, aiming to profile transcriptome, proteome, and sometimes epigenome from a single tissue section rather than inferring one layer from another. A recent review of single-cell and spatial multiomics methods described the computational integration challenge this creates as one of the field's defining bottlenecks, since combining multiple molecular layers from the same tissue rarely produces cleanly aligned datasets.

Some multiomics approaches co-detect RNA and protein from the same section directly, while others rely on registering separate spatial transcriptomics and spatial proteomics datasets from adjacent tissue slices after the fact. Each strategy trades off differently: co-detection avoids alignment error but currently supports fewer targets, while serial-section registration scales to more modalities but introduces the risk that adjacent slices do not perfectly represent the same cellular neighborhood.

Deciding among these three layers rarely comes down to a single deciding factor. A project studying gene regulatory programs in development will likely start with transcriptomics, a project validating a specific protein target for a companion diagnostic will lean toward proteomics, and a project trying to explain why two cells with similar transcriptomes behave differently will eventually need both, which is where multiomics approaches become necessary rather than optional.

Spatial biology technologies and platforms

Sequencing-based spatial transcriptomics platforms capture RNA using spatially barcoded arrays, then reconstruct gene expression location computationally. Their defining trade-off is spot size vs coverage: earlier platforms captured whole-transcriptome data from spots containing several cells, while newer high-density arrays are closing in on single-cell resolution without sacrificing gene coverage.

Imaging-based platforms take the opposite approach, using iterative rounds of fluorescence hybridization or in situ sequencing to localize individual RNA molecules at subcellular resolution. A comparison of imaging platforms tested in formalin-fixed, paraffin-embedded (FFPE) tumor samples found meaningful differences in sensitivity and cell segmentation performance depending on tissue type and platform, underscoring that no single platform is universally superior.

On the proteomics side, cyclic immunofluorescence methods use standard fluorescence microscopes and repeated stain-image-bleach cycles, while metal-tag approaches such as IMC avoid fluorescence's spectral overlap entirely by reading antibodies through mass spectrometry instead. A recent review of translational spatial proteomics notes that cyclic immunofluorescence, co-detection by indexing, and IMC collectively now enable spatial localization of more than 50 proteins at subcellular resolution.

A fourth technology family works from the analytical chemistry side rather than antibodies or sequencing, mapping metabolites, lipids, and some proteins directly from tissue without labels of any kind. Mass spectrometry-based spatial analysis is the only spatial modality that can access the metabolic layer transcriptomics and antibody-based proteomics both miss, and it is treated separately from the platforms above given its distinct instrumentation and analytical chemistry roots.

Vendor competition across these categories has accelerated rapidly, with commercial platforms from companies such as 10x Genomics, Bruker, Akoya Biosciences, Vizgen, Standard BioTools, Ionpath, Leica, and Zeiss each occupying a different point on the resolution, plex, and throughput spectrum. No single vendor currently leads across every category, which is part of why platform selection has become its own specialized decision rather than a default choice.

Instrument footprint, sample throughput, and tissue compatibility vary just as widely as the underlying chemistry does. A platform that performs well on fresh frozen tissue may require substantial optimization for FFPE samples, and researchers evaluating a new platform for the first time should expect a validation period before generating publication-ready data from their own tissue types.

Cost has fallen steadily as these platforms have matured, though it remains a significant factor in project planning. Consumable costs per sample, instrument access fees at shared core facilities, and the downstream computational cost of storing and analyzing the resulting data all belong in an honest budget for a first spatial biology project, alongside the instrument or service fees that tend to get the most attention up front.

Table 1: A comparison of the major spatial biology modalities and their typical resolution and readout.

Modality	Primary readout	Typical resolution	Example platform types
Sequencing-based transcriptomics	Whole transcriptome	Spot to near single-cell	Barcoded capture arrays
Imaging-based transcriptomics	Targeted RNA panel	Subcellular	Multiplexed fluorescence hybridization
Spatial proteomics	Targeted protein panel	Subcellular	Cyclic fluorescence, metal-tag imaging
Spatial multiomics	Two or more molecular layers	Varies by layer	Co-detection, serial-section methods

How spatial biology data is analyzed

Generating a spatial dataset is only the first half of the work. Imaging-based platforms in particular require cell segmentation, the process of drawing accurate cell boundaries in tissue, before any downstream biology can be extracted, and segmentation errors propagate into every analysis that follows.

Once cells are segmented, sequencing-based spot data with multiple cells per spot typically requires deconvolution, a computational step that infers cell-type composition within each spot using a matched single-cell reference dataset. Imaging-based data, by contrast, moves more directly to cell typing, using marker genes or reference mapping to assign identities to already-segmented cells.

A recent benchmarking of imaging-based workflows recommended best-practice approaches for preprocessing, segmentation, and spatially variable feature selection, noting that analysis choices at each step materially affect the biological conclusions researchers draw from the same raw dataset. Open-source toolkits such as Squidpy, Seurat, and Giotto have become the standard software layer for this work, though which toolkit fits best depends heavily on whether a lab's broader pipeline runs in Python or R.

Figure 1: A linear flowchart of the spatial biology data pipeline, from raw tissue image or sequencing output through segmentation, cell typing, and spatial statistics to biological interpretation. Credit: AI-generated image created using Google Gemini (2026).

The final analytical step, once cells are typed and mapped, is asking what the spatial arrangement itself reveals: which cell types sit near each other more often than chance would predict, and what that proximity implies about signaling or disease progression. This is the layer of analysis unique to spatial data, and the workflows, tools, and best practices for that step are substantial enough to warrant their own dedicated treatment.

Machine learning is increasingly embedded throughout this pipeline rather than applied only at the end. Deep-learning segmentation models have replaced classical image-processing approaches in many labs, and unsupervised clustering methods are now standard for identifying spatial domains or niches that were not defined in advance by the researcher.

Data volume compounds every one of these challenges. A single spatial imaging run can generate gigapixel-scale image files alongside molecular readouts for hundreds of thousands of cells, and pipelines built for smaller single-cell datasets often need meaningful re-engineering before they scale to a typical spatial experiment.

Reproducibility is a persistent concern across the analysis stack, since small differences in segmentation parameters, normalization approach, or reference dataset can shift downstream cell-type calls in ways that are easy to overlook. Labs that document and version their analysis pipelines alongside their wet-lab protocols tend to have an easier time reconciling results across experiments, instruments, and personnel over time.

Applications of spatial biology across research fields

Cancer research has adopted spatial biology faster than almost any other field, largely because tumor microenvironment biology is inherently spatial. Whether immune cells can physically reach and infiltrate a tumor, and what signals surrounding stromal cells send in response, are questions that dissociated sequencing simply cannot answer, and spatial platforms have already been used to map how cancer and immune cell communication patterns relate to a patient's response to immunotherapy.

Neuroscience has embraced spatial transcriptomics for a related reason: the brain's function is inseparable from its precise cytoarchitecture, and dissociating brain tissue destroys the regional organization that spatial methods were built to preserve. Large-scale brain atlas projects are now using spatial data to map cell types onto anatomical regions at a resolution that was not possible even 5 years ago.

Immunology and developmental biology have followed a similar path, using spatial data to study immune cell niches, tertiary lymphoid structures, and the choreography of gene expression during embryogenesis and organogenesis. Applications across these research fields continue to expand as platform costs fall and analysis pipelines mature, with infectious disease and microbiome research now emerging as the next frontier.

Pharmaceutical research has taken up spatial biology as well, using it to find and validate drug targets in native tissue context rather than in dissociated cells. A growing body of work on spatial omics in oncology has positioned the approach as a tool for target discovery and translational biomarker development, a use case distinct from, but complementary to, the basic-research applications covered above.

Large reference atlas projects have become one of the field's most visible applications, using spatial data alongside single-cell references to build increasingly detailed maps of human tissue across organs, developmental stages, and disease states. These atlases are becoming a shared resource that individual labs draw on for cell-type annotation rather than something any single group could build alone.

Precision medicine is an emerging application that ties several of these fields together. Spatial signatures that combine cell composition, spatial arrangement, and molecular state are being explored as predictors of therapy response in settings where a single biomarker has proven insufficient, particularly in immuno-oncology, where the physical arrangement of immune and tumor cells appears to carry information that expression levels alone do not.

Infectious disease research and microbiome studies represent the newest expansion of the field, applying the same spatial logic to host-pathogen interactions and to how microbial communities organize within a tissue. These applications remain earlier stage than the cancer and neuroscience work described above, but they illustrate how quickly a technique developed for one research question tends to migrate into others once the platforms and analysis tools mature.

Choosing the right spatial biology approach

Selecting a spatial method starts with the biological question, not the platform's marketing claims. A discovery-stage project with no strong hypothesis about which genes matter benefits from whole-transcriptome, sequencing-based methods, while a project built around a defined set of markers is usually better served by a targeted, high-resolution imaging platform.

The following framework covers the core decisions most researchers face when planning a spatial biology project:

1. Define whether the question requires unbiased, genome-wide discovery or a targeted, high-resolution readout of known markers.
2. Decide whether RNA, protein, or both layers are needed to answer the question, since spatial proteomics and spatial transcriptomics require different sample preparation and instrumentation.
3. Assess tissue compatibility, including whether samples are fresh frozen or FFPE, since this affects which platforms are viable.
4. Confirm that computational resources and staff expertise exist for the segmentation, cell typing, and spatial statistics steps that follow data generation.
5. Weigh throughput and cost against the resolution and plex the biology actually demands, rather than defaulting to the highest-specification platform available.

The operational and budget side of that decision, including staffing, procurement, and total cost of ownership, is a distinct discipline from the science covered here and is addressed at length for lab managers evaluating, implementing, and scaling spatial technologies. For projects centered on metabolites and lipids rather than nucleic acids or proteins, the decision runs through mass spectrometry imaging rather than any of the platforms discussed above, an analytical chemistry approach covered from the practitioner's side elsewhere.

Few individual labs have every skill this decision requires in-house, from histology and sample prep through instrument operation to computational analysis. Many groups start by outsourcing a first spatial experiment to a core facility or commercial service provider, then decide whether project volume justifies bringing a platform in-house once the approach has been validated on their own biology.

Reassessing the choice is normal rather than a sign of a poor initial decision. As platforms mature, resolution improves, gene panels expand, and new co-detection chemistries arrive, a method that was the right fit 18 months ago may no longer be the best available option, which is why most groups treat platform selection as an ongoing decision rather than a one-time choice.

Spatial biology's foundation for future research

Spatial biology has moved from a specialized technique to a standard part of the modern research toolkit, and the pace of platform development shows no sign of slowing. Researchers evaluating a spatial project today have more method choices, more mature analysis software, and more published benchmarking data to draw on than at any previous point in the field's history.

The throughline across transcriptomics, proteomics, and multiomics methods is the same: location is data, and discarding it during dissociation throws away information that often explains why a treatment works, why a cell state persists, or why a tissue fails to repair itself. That principle is likely to keep expanding spatial biology's reach into new fields well beyond the cancer and neuroscience research where it first took hold.

For researchers deciding where to start, the practical path is rarely to master every method described here at once. It is to identify the specific molecular layer, resolution, and analysis capability a given question actually requires, and to build outward from that first well-scoped project as the platform landscape and internal expertise both continue to mature.

This content includes text that has been created with the assistance of generative AI and has undergone editorial review before publishing. Technology Networks' AI policy can be found here.