For decades, drug discovery has relied heavily on molecular measurements averaged across thousands or millions of cells. These approaches remain valuable, but an average can conceal the differences that matter most: a rare resistant population, a transient disease-associated state, or a small group of cells responding differently to treatment.
Single-cell technologies have changed this view by allowing tissues, tumors, and experimental models to be studied as dynamic cellular ecosystems rather than uniform samples. Cell-atlas initiatives have created increasingly detailed maps of human cell types and states, while single-cell studies are beginning to connect this diversity with key stages of drug discovery. The opportunity is, therefore, not simply to describe more cell populations; it is to identify which states drive disease, which can be altered, and which should influence the selection of targets, models, biomarkers, and treatment strategies. Moving from maps to decisions is where single-cell research has its greatest impact.
Where single-cell insight changes drug discovery decisions
The value of single-cell research lies not only in identifying previously unrecognized cell populations, but in determining how cellular diversity should change the direction of a discovery program. By locating a potential target within specific disease-associated cell types or states, researchers can assess whether it is likely to act on the intended biology, which patient populations may be most relevant, and where unwanted effects could arise.
This perspective shapes decisions throughout drug discovery. At the target-identification stage, single-cell analysis can distinguish cells actively driving pathology from neighboring cells that may appear similar in bulk data. In models of inflammatory arthritis, for example, researchers identified functionally distinct fibroblast populations associated separately with inflammation and tissue damage, illustrating how cellular resolution reveals more precise therapeutic opportunities.
Single-cell approaches can also help researchers select more representative experimental models, develop biomarker hypotheses, and examine why responses differ among cells exposed to the same treatment. Rather than categorizing an entire sample as responsive or resistant, researchers can identify populations that disappear, persist, adapt, or emerge following intervention. Recent studies have even explored whether single-cell tumor transcriptomes can support predictions of treatment response and resistance.
Moving from observation to function
Identifying a disease-associated cell state is an important starting point, but it does not establish what controls that state or whether it can be therapeutically altered.
Genetic perturbations, compound treatments, longitudinal sampling, and advanced models such as organoids can help researchers examine how individual cell populations respond when a pathway is disrupted, or a candidate therapy is introduced. This can reveal whether an intervention produces the intended molecular effect, whether different cell types respond in distinct ways, and whether small populations persist or adapt following treatment.
Large-scale studies illustrate what is becoming possible. Using sci-Plex, researchers profiled approximately 650,000 single-cell transcriptomes from three cancer cell lines exposed to 188 compounds, while genome-scale Perturb-seq connected CRISPR interference with transcriptional phenotypes across more than 2.5 million cells.
Not every discovery program requires screening at this scale. The broader principle is to move beyond observing which cells are associated with disease and test how those cells behave when their underlying biology is perturbed. This progression from observation to functional evidence can strengthen target validation, clarify mechanism of action, and reveal potential routes to resistance earlier in discovery.
Building a more complete view through spatial and multiomics research
Single-cell transcriptomics can reveal which cells and molecular states are present, but cellular identity alone does not provide the complete biological picture. Dissociating a tissue removes information about where cells were located, which populations surrounded them, and how local interactions may have influenced their behavior. Transcript abundance also does not always explain the genomic, epigenetic, protein, or metabolic processes underlying an observed cell state.
Integrating single-cell data with spatial and complementary omics approaches begins to address these gaps. Spatial analysis can reveal tumor boundaries, immune-excluded regions, cellular neighborhoods, and interactions among malignant, immune, and stromal populations. In oral squamous cell carcinoma, for example, an integrated single-cell and spatial transcriptomic study identified distinct tumor-core and leading-edge architectures associated with survival and predicted response to targeted therapy.
Multiomics research can add further layers of functional evidence. A study combining single-cell RNA sequencing, spatial transcriptomics, spatial proteomics, and perturbation experiments identified distinct cellular ecosystems and an immunosuppressive signaling interaction in cervical cancer.
Greater resolution does not automatically mean better decisions
As single-cell technologies become more accessible, it can be tempting to equate larger datasets, more cell clusters, or greater analytical complexity with better scientific insight. Yet the usefulness of a study ultimately depends on whether its design can answer the biological and drug discovery questions that prompted it.
The number of cells required, for example, should reflect the expected heterogeneity of the sample, the rarity of the populations of interest, and the comparisons researchers need to make. Biological replication, representative models, appropriate treatment conditions, and carefully selected time points may be more important than maximizing cell numbers. Sample collection and processing must also be planned carefully, as technical variation can become difficult or impossible to distinguish from genuine biology when experimental groups are confounded across batches. Published guidelines for single-cell experimental design emphasize the need to consider sample preparation, sequencing, and analysis together from the outset.
Among service laboratories running single-cell and multiomics studies at scale, projects tend to produce more actionable results when the biological question is defined before the platform is selected. Reversing that order can be costly: a study may be sized to a budget rather than to the rarity of the population it is intended to detect, or treatment groups may be split across sequencing batches, making technical and biological variation difficult to separate. Defining the question first helps turn computationally identified populations into testable findings rather than additional clusters on a plot.
Integrated and decision-focused single-cell research
The next phase of single-cell research will not be defined solely by the number of cells profiled or the resolution of new atlases. Its influence on drug discovery will depend on how effectively cellular diversity can be connected with mechanism, function, tissue context, and therapeutic response. AI-based approaches may help integrate complex datasets and identify patterns that warrant further biological investigation.
The real opportunity is not simply to see biology in greater detail, but to use that detail to reduce uncertainty and guide better decisions across the discovery process.
