The dawn of an era

On January 6, 2021, Nature Methods unanimously declared spatial transcriptomics “Method of the Year” [1]. The reason was clear: single-cell sequencing had been revealing cellular heterogeneity, but we are three-dimensional beings, and knowing how cells are physically arranged is critical for improving biological models, frameworks, and drug therapies.

Cancer research shows why this is transformative [2]. A ground-up tumor sample processed for NGS sequencing can reveal mutations and expression patterns, but may not reveal the degree of immune cell infiltration. Although such information can be critical for drug therapy selection strategies. Staining can show infiltration but not the subtypes of T cells or B cells present in the sample, which may also be critical to guide the drug therapy selection. Single-cell RNA-seq can provide this subtype detail, but destroys spatial context. Collecting all this separately is either cumbersome or impossible from one biopsy.

Spatial transcriptomics overcomes this hurdle. From a single section, researchers can see structure, cell types, immune infiltration, and interactions, yielding a high-resolution map of the tumor microenvironment that was previously unattainable.

From bleeding edge to leading edge

Spatial transcriptomics is split into two areas: sequencing-based and imaging-based.

Sequencing-based approaches can capture any polyadenylated transcript, thus measuring most of the protein-coding genes within the genome, and are therefore sequence-agnostic [3]. However, their main drawback is resolution. Although the state-of-the-art, Visium HD, improved this resolution, jumping from 55 μm spot sizes to 2 μm squares, it remains less precise than imaging methods.

Imaging approaches, such as MERFISH [4] and RNAscope [5], extended single-molecule FISH (smFISH), which dates to 2008 [6], though in situ hybridization itself was described decades earlier [7]. These methods offered subcellular precision but were initially limited to only a handful of targets, requiring prior knowledge of which genes to probe [2]. Validation was extensive, and experiments were time-intensive [8]. Still, their spatial accuracy surpassed sequencing methods [8,9].

Both approaches carried trade-offs, but today imaging-based methods, particularly NanoString’s CosMx and 10x Genomics’ Xenium, are the dominant platforms. They now offer panels of 5,000–6,000 genes. While short of ~20,000 protein-coding genes, in many instances, this is enough to capture the most important biologically relevant signals. Their specificity is also high, with reported single-molecule detection sensitivities above 90–95% for many targets and low false-positive rates. Crucially, they retain subcellular resolution. So, Imaging-based transcriptomics has moved past its breakthrough stage and has very much arrived.

So many FISH in the sea

Modern imaging-based spatial transcriptomic methods improved on their smFISH predecessor through greater signal amplification and clever gene decoding strategies [2]. A single fluorophore is too weak, so primary probes are amplified through recruiting numerous secondary fluorophore-bound probes. Decoding over successive imaging cycles allows for measuring thousands of transcripts simultaneously despite the limits of modern fluorophore spectra. 

Xenium uses padlock probes, which circularize upon binding RNA, enabling rolling circle amplification to generate long concatemeric DNA products with many secondary probe binding sites [10,11]. Each transcript is decoded through successive cycles of fluorescent probe binding, with built-in redundancies to reduce errors. Xenium panels cover ~5,000 genes plus ~100 custom targets, through the addition of antibodies or stains, can provide multi-omic data as well.

CosMx uses a different strategy involving DNA probes containing multiple readout domains, without rolling circle amplification or a padlock design. Instead, signal amplification occurs through secondary probes arranged in a branched configuration, each carrying numerous fluorophores. In each imaging cycle, these branched fluorescent reporters bind to the readout domains. After imaging, the fluorophores are cleaved off, leaving the probes intact for subsequent imaging rounds [4]. Multiple probes per transcript provide redundancy. CosMx panels currently target ~6,000 genes plus ~200 custom genes.

Other commercial platforms include seqFISH [12] (Spatial Genomics) and Molecular Cartography [13] (Resolve Biosciences).

And then there were two

Currently, Xenium and CosMx, while differing in chemistry, probe counts, and workflows, are the current commercial front-runners, so it’s relevant to briefly compare their utility. 

Xenium is often seen as more reliable for “ground truth,” producing consistent transcript counts with low false positives. CosMx is favored for exploratory work, thanks to panel flexibility and early protein integration. Yet performance is gene-dependent; even if one platform is generally better, it may underperform on certain transcripts [14,15].

Although the rapid developments make comparing these platforms challenging, there is some recent research we can look to. One study found Xenium yielded higher transcript counts in FFPE tissues while maintaining specificity [14]; another reported its custom 290-plex Xenium gene panel showed the highest sensitivity and specificity [16]. A recent Nature Methods benchmark of seven platforms confirmed Xenium’s improved detection efficiency [15].

Current successes and future promises

The benefits are already visible. Spatial transcriptomics showed human brain structures are pre-patterned at the molecular level before being anatomically distinct, overturning assumptions about cortical gene expression [17]. In breast cancer, it highlighted tumor heterogeneity with potential clinical impact [18]. In colorectal cancer, it revealed macrophage subpopulations that act as pro- or anti-tumor depending on context [19]. Notably, these studies used smaller and now outdated gene panels, thus highlighting a lot of untapped potential for future research.

The coming years will bring deeper insights into development, disease, and therapeutic response. Perhaps most promising is clinical integration. Analyzing patient biopsies with spatial transcriptomics could guide treatment decisions and drug development. While its full impact remains ahead, spatial omics is poised to reshape biology and medicine.

Sean O’Toole

Reference

1. Method of the Year 2020: spatially resolved transcriptomics. Nat. Methods 18, 1–1 (2021).
2. Jin, Y. et al. Advances in spatial transcriptomics and its applications in cancer research. Mol. Cancer 23, 129 (2024).
3. Ståhl, P. L. et al. Visualization and analysis of gene expression in tissue sections by spatial transcriptomics. Science 353, 78–82 (2016).
4. Chen, K. H., Boettiger, A. N., Moffitt, J. R., Wang, S. & Zhuang, X. Spatially resolved, highly multiplexed RNA profiling in single cells. Science 348, aaa6090 (2015).
5. Wang, F. et al. RNAscope: A Novel in Situ RNA Analysis Platform for Formalin-Fixed, Paraffin-Embedded Tissues. J. Mol. Diagn. 14, 22–29 (2012).
6. Raj, A., van den Bogaard, P., Rifkin, S. A., van Oudenaarden, A. & Tyagi, S. Imaging individual mRNA molecules using multiple singly labeled probes. Nat. Methods 5, 877–879 (2008).
7. Gall, J. G. & Pardue, M. L. Formation and detection of rna-dna hybrid molecules in cytological preparations*. Proc. Natl. Acad. Sci. 63, 378–383 (1969).
8. Strell, C. et al. Placing RNA in context and space – methods for spatially resolved transcriptomics. FEBS J. 286, 1468–1481 (2019).
9. Linares, A., Brighi, C., Espinola, S., Bacchi, F. & Crevenna, Á. H. Structured Illumination Microscopy Improves Spot Detection Performance in Spatial Transcriptomics. Cells 12, 1310 (2023).
10. Nilsson, M. et al. Padlock Probes: Circularizing Oligonucleotides for Localized DNA Detection. Science 265, 2085–2088 (1994).
11. Lizardi, P. M. et al. Mutation detection and single-molecule counting using isothermal rolling-circle amplification. Nat. Genet. 19, 225–232 (1998).
12. Eng, C.-H. L. et al. Transcriptome-scale super-resolved imaging in tissues by RNA seqFISH+. Nature 568, 235–239 (2019).
13. Alon, S. et al. Expansion sequencing: Spatially precise in situ transcriptomics in intact biological systems. Science 371, eaax2656 (2021).
14. Wang, H. et al. Systematic benchmarking of imaging spatial transcriptomics platforms in FFPE tissues. bioRxiv 2023.12.07.570603 (2023) doi:10.1101/2023.12.07.570603.
15. Marco Salas, S. et al. Optimizing Xenium In Situ data utility by quality assessment and best-practice analysis workflows. Nat. Methods 22, 813–823 (2025).
16. Mennillo, E. et al. Single-cell spatial transcriptomics of fixed, paraffin-embedded biopsies reveals colitis-associated cell networks. bioRxiv 2024.11.11.623014 (2024) doi:10.1101/2024.11.11.623014.
17. Qian, X. et al. Spatial transcriptomics reveals human cortical layer and area specification. Nature 644, 153–163 (2025).
18. Wang, X. et al. Spatial transcriptomics reveals substantial heterogeneity in triple-negative breast cancer with potential clinical implications. Nat. Commun. 15, 10232 (2024).
19. Oliveira, M. F. de et al. High-definition spatial transcriptomic profiling of immune cell populations in colorectal cancer. Nat. Genet.57, 1512–1523 (2025).