How do you find drug targets with spatial biology?

By measuring gene or protein expression while preserving each cell’s tissue location, then examining two properties bulk data cannot separate: whether expression concentrates in a specific cell type, and whether that cell type’s relevance depends on its physical position relative to other tissue structures. Both properties can inform which candidates are prioritized for further validation.

What is a cell-type-specific target?

A target whose expression concentrates in a defined cell population rather than spreading diffusely across many cell types in a tissue sample. This property is visible in dissociated single-cell data, though spatial measurement can confirm it is not distorted by cell types that survive tissue dissociation differently than others.

Why does target location matter?

Because a cell’s behavior, and therefore a target’s therapeutic relevance, can depend on where that cell sits relative to a disease structure. Research on Alzheimer’s disease found that microglia near amyloid plaques adopt a distinct immunometabolic state driven specifically by proximity, a distinction no dissociated single-cell assay can detect since it destroys positional information entirely.

Finding drug targets in tissue context

Highlights in this article

The limits of bulk target discovery
Cell-type-specific targets
Location-dependent target biology
Case examples
Integrating with existing target pipelines

Getting spatial target identification right starts with a distinction that bulk and dissociated single-cell approaches cannot make: whether a candidate gene matters because of what it is, or because of where it sits. Both properties can point to the same gene, but only one of them is visible without spatial data, and it is often the one that decides whether a target actually works as a therapeutic hypothesis.


Key takeaways Bulk sequencing reports an average across a mixed sample, which can hide a target that matters intensely in one cell type or tissue region and not at all elsewhere. A cell-type-specific target is defined by which cells express it. A location-dependent target is defined by where those cells sit relative to everything around them, a distinct and separate property. In Alzheimer’s research, a set of genes specifically co-localizing with amyloid plaques revealed that microglial behavior changes with physical proximity to plaque pathology, not with cell identity alone. In alcohol-related liver disease, mapping where fibrosis-associated cells concentrated in a cirrhotic liver led directly to identifying a bulk blood biomarker of disease severity. Spatial evidence works best layered onto an existing target pipeline built on genetics and bulk omics, not as a replacement for it.

Key takeaways

Bulk sequencing reports an average across a mixed sample, which can hide a target that matters intensely in one cell type or tissue region and not at all elsewhere.
A cell-type-specific target is defined by which cells express it. A location-dependent target is defined by where those cells sit relative to everything around them, a distinct and separate property.
In Alzheimer’s research, a set of genes specifically co-localizing with amyloid plaques revealed that microglial behavior changes with physical proximity to plaque pathology, not with cell identity alone.
In alcohol-related liver disease, mapping where fibrosis-associated cells concentrated in a cirrhotic liver led directly to identifying a bulk blood biomarker of disease severity.
Spatial evidence works best layered onto an existing target pipeline built on genetics and bulk omics, not as a replacement for it.

The limits of bulk target discovery

A bulk RNA sequencing measurement reports a single expression value for an entire tissue sample, effectively averaging across every cell type present in whatever proportion they happened to occur. That average is genuinely useful for many questions, but it is structurally incapable of answering two questions that matter enormously for target discovery: which specific cell type is responsible for an expression signal, and whether that cell type’s behavior depends on its physical position within the tissue.

Single-cell RNA sequencing solved the first problem by measuring gene expression in individual dissociated cells, which is a major advance over bulk data. It does not solve the second, because dissociating tissue into a cell suspension necessarily destroys the spatial relationships between cells. A cell type identified as expressing a candidate target in single-cell data still cannot tell a researcher whether that cell, in the intact tissue, sat adjacent to an immune infiltrate, near a blood vessel, or isolated in an otherwise unremarkable region, because that information no longer exists once the tissue has been dissociated.

Spatial transcriptomics and spatial proteomics close that specific gap by measuring expression while keeping every cell’s tissue coordinates intact. That is the single capability this entire spoke is built around, and it splits into two distinct target properties worth treating separately, since they require different kinds of evidence and have different implications for a discovery program.

Cell-type-specific targets

A cell-type-specific target is one whose expression concentrates in a defined cell population rather than spreading diffusely across many cell types in a tissue. This is the property single-cell data can already reveal without spatial resolution, and it remains foundational: a target expressed narrowly in a disease-relevant cell type is generally a cleaner therapeutic hypothesis than one expressed broadly, since narrower expression usually implies fewer off-target effects from modulating it.

What spatial data adds to this property specifically is confirmation and context rather than a wholly new capability. Spatial measurement can confirm that a cell-type-specific signal identified in dissociated data is not a processing artifact of dissociation itself, since some cell types survive tissue dissociation better than others, which can distort apparent cell-type proportions and, in turn, distort which expression signals look concentrated versus diffuse. Measuring expression directly in intact tissue removes that specific source of error.

A kidney disease example illustrates why this distinction matters practically. The kidney contains more than 20 distinct cell types in its mature state, and mapping regional gene expression patterns and intercellular communication networks across those cell types, rather than relying on dissociated proportions alone, has directly supported identifying therapeutic targets in diabetic kidney disease. A complex, multi-cell-type organ of this kind is precisely where dissociation-related distortion is most likely to mislead a cell-type-specificity claim, and where spatial confirmation earns its cost.

Location-dependent target biology

This is the property bulk and dissociated single-cell data genuinely cannot measure at all, not merely measure imperfectly. A location-dependent target is one whose relevance depends specifically on where its expressing cells sit relative to other structures in the tissue, a property that has no dissociated analog because dissociation eliminates position entirely.

The clearest evidence for why this matters comes from neurodegeneration research. Work on Alzheimer’s disease identified a network of co-expressed genes, since termed Plaque Induced Genes, that specifically co-localize with amyloid plaques, with both the connectivity and expression level of this gene network increasing as plaque burden increases. The genes are microglial and astrocytic in origin, and they implicate the complement system, oxidative stress, lysosomal function, and inflammation, a specific molecular signature that only appears where plaques physically are.


When proximity itself is the biology Follow-up work using higher-resolution spatial profiling technologies went further, identifying distinct cellular neighborhoods in which activated microglia specifically co-localize with reactive astrocytes and other glial cells around pathology, describing specific niches of neuroinflammatory metabolism. The researchers found that anatomical context and disease environment together shape microglial immunometabolic states, meaning the same microglial cell type behaves differently depending on its physical distance from plaque pathology. That is location-dependent target biology stated as precisely as the evidence allows: not a claim that microglia matter in Alzheimer’s, which single-cell data had already shown, but a claim that which microglia matter, and how, depends on a spatial relationship no dissociated assay can capture. The researchers describe this as filling a gap between single-cell data and histopathology and laying groundwork for spatially targeted therapeutic strategies, which is the practical payoff a target discovery team should be looking for from this kind of finding.

When proximity itself is the biology

Follow-up work using higher-resolution spatial profiling technologies went further, identifying distinct cellular neighborhoods in which activated microglia specifically co-localize with reactive astrocytes and other glial cells around pathology, describing specific niches of neuroinflammatory metabolism. The researchers found that anatomical context and disease environment together shape microglial immunometabolic states, meaning the same microglial cell type behaves differently depending on its physical distance from plaque pathology.

That is location-dependent target biology stated as precisely as the evidence allows: not a claim that microglia matter in Alzheimer’s, which single-cell data had already shown, but a claim that which microglia matter, and how, depends on a spatial relationship no dissociated assay can capture. The researchers describe this as filling a gap between single-cell data and histopathology and laying groundwork for spatially targeted therapeutic strategies, which is the practical payoff a target discovery team should be looking for from this kind of finding.

Case examples

Beyond the Alzheimer’s example above, two further cases show how this evidence has translated into concrete targets or biomarkers across different disease areas, deliberately outside oncology, where this cluster’s other articles already provide extensive case material.

Disease area	Spatial finding	What it produced
Alzheimer’s disease	A gene network co-localizing specifically with amyloid plaques, with expression and connectivity scaling with plaque burden; microglia near plaques show a distinct immunometabolic state driven by proximity	A spatially defined target hypothesis centered on the microglia-astrocyte plaque niche rather than microglia generally
Alcohol-related liver disease	Ultra-high-resolution spatial transcriptomics of cirrhotic liver found hepatic stellate cells concentrated specifically in fibrotic zones, with fibrotic areas showing reduced transcriptomic accessibility overall	Identification of systemic soluble TREM2 as a blood-measurable indicator of liver disease severity
Diabetic kidney disease	Regional gene expression and intercellular communication mapped across the kidney’s more than 20 cell types	Support for therapeutic target identification specific to disease-relevant renal cell populations

Table 1. Three disease areas where spatial evidence produced a specific target or biomarker outcome, chosen to span neurodegeneration, liver disease, and kidney disease rather than repeating an oncology example.

The liver disease case is worth expanding because it demonstrates the full pipeline this spoke describes, from a spatial observation to a usable clinical readout. Researchers applying ultra-high-resolution spatial transcriptomics, at a resolution of 2 microns per measurement spot, to cirrhotic liver tissue from a patient with end-stage alcohol-related liver disease found that hepatic stellate cells, the cell type most directly responsible for fibrosis, were predominantly located within fibrotic areas rather than distributed through the parenchyma generally, and that fibrotic regions themselves contained fewer accessible measurement spots than non-scarred tissue. That spatial pattern of cellular concentration led the researchers to identify systemic soluble TREM2 levels as an indicator of the degree of liver disease, translating a location-specific cellular observation into a biomarker that does not itself require spatial measurement to use clinically.

That translation, from a spatially resolved cellular observation to a bulk, clinically actionable readout, is the throughline connecting all three cases in the table. A finding does not have to remain a spatial dataset forever to be useful. It has to start as one.

Integrating with existing target pipelines

Spatial evidence is most useful layered onto a target discovery pipeline that already runs on genetic association data and bulk expression, not as a wholesale replacement for either. Four integration points make that layering practical.

Use spatial data to disambiguate genetic hits. A genome-wide association signal near a candidate gene does not specify which cell type, or which tissue location, is actually responsible for the disease association. Spatial expression data can narrow that ambiguity directly.
Use spatial data to re-examine bulk omics results that failed to translate. A target that looked promising in bulk tissue data but failed in later validation is worth re-examining spatially, since the original bulk signal may have reflected an average obscuring a genuinely strong but spatially concentrated effect.
Weight spatial and single-cell tissue-context evidence alongside genetic evidence when prioritizing candidates. Tissue-context expression evidence has been shown to identify a target space that is broader than, and complementary to, genetic evidence alone, which argues for using both rather than treating either as sufficient on its own.
Reserve full spatial profiling for candidates that have already cleared an initial prioritization step. Spatial experiments remain more resource-intensive than bulk or single-cell approaches, so the practical sequence is usually genetics and bulk data first, spatial confirmation on a narrowed candidate list second.

Combining spatial data with other omics layers on the same tissue, rather than using spatial transcriptomics alone, is where several of the strongest recent findings, including the Alzheimer’s and liver examples above, have actually come from. What that combined data shows in practice, and how to build an analysis around more than one spatial modality at once, is covered in Spatial Multi-Omics for Target Discovery: What the Data Actually Shows.

For the broader tissue-context argument this spoke develops, including the clinical-success evidence behind it, see Spatial biology for target discovery: Mapping disease at tissue resolution, and for where target discovery sits within the full spatial biology pipeline, Spatial biology in drug discovery: From target discovery to translational medicine.

This article was produced under Drug Discovery News’s editorial policies.