Ask the question your last study could not.
The observation you could not explain in one study may become the question that starts the next. Your existing slides give you a way to pursue it, including cases that were never molecularly profiled.
Aurora predicts spatial gene expression from H&E images, bringing genes into the same view as the tissue pattern that caught your attention. You can follow the idea across a cohort or beyond a targeted panel, using earlier work to develop a new line of research.
Applications
Turn unfinished questions into new lines of research
- Explore
The idea that arrived after the study ended
When a new hypothesis brings you back to a published cohort, predicted expression lets you investigate it in cases the original experiments never profiled. Even a sample whose block has been exhausted can contribute through its digitised slide.
Built onOut-of-the-box predictionComparative spatial analysis
DeepSpot-M predicts 19,338 protein-coding genes at every retained tile of a slide, from the H&E image alone.[1]
- Explore
The observation that stayed with you
A boundary or region that stood out under the microscope gives you a concrete place to start. By examining predicted expression there and in the surrounding tissue, you can turn that observation into a question about which genes vary and where.
Which genes vary most across the section?
Spatially variable genes, ranked by Moran's I, in the spatial analysis report
Which regions carry different expression programmes?
Clusters drawn on the tissue, with their marker genes, in the spatial analysis report
Which clusters sit next to each other in the tissue?
Neighbourhood enrichment between clusters, in the spatial analysis report
How does one gene look on the tissue, and on the UMAP?
Any predicted gene, on the tissue and on the UMAP, in the explorer, on the Academic Research route
What separates one region from another?
Differential expression between two groups you select, in the explorer, on the Academic Research route
- Extend
The result that pointed past the panel
An unexpected result can lead to genes you had no reason to include when you designed the panel. With our team, you can adapt the model using your measured examples to explore additional predicted genes in the same regions, or revisit selected genes lost during quality control.
Built onPanel extensionRestore genesFinetune
Arranged with our team, on the Academic Research route.Learn about Academic Research
The DeepSpot-M preprint demonstrates prediction for genes excluded from training.[1]
- Look ahead
The question left open for the next study
A proposal becomes more specific when you can identify the pattern you want to investigate and the material in which to study it. Predicted maps help you develop that connection across existing slides before selecting samples and regions for the next experiment.
Your route
Discuss your research idea
Bring a new question to an existing cohort, or tell us about the study you want to build next.
University groups, institutes, hospitals and foundations working on their own cohorts.
Submit an H&E slide and read predicted spatial gene expression back, with no assay to book and no agreement to sign. Analysis is for academic and non-profit research. If your work sits between academic and commercial, ask before you prepare a slide.
Worked examples
Follow an idea from an existing study

Give your archive a new role in discovery.
Bring unprofiled cases into molecular research and look for shared patterns across your archive, using the slides you already hold.
- You bring
- A digitised H&E slide for each sample you want to explore.
Follow discoveries beyond your original panel.
Follow a finding beyond the genes you originally selected, with additional predictions in the tissue regions your experiment already measured.
Arranged with our team
- You bring
- An H&E image and measured spatial expression, with coordinates for each spot.

Shape your next experiment with a wider view.
Build your study around the patterns you want to investigate, with a view of candidate samples and tissue regions before the assay begins.
- You bring
- The slides you are choosing between.
References
- [1]Nonchev K, Dawo S, Silina K, Koelzer VH, Rätsch G. DeepSpot-M: a multimodal foundation model for transcriptome-wide virtual spatial transcriptomics from histology. medRxiv. Preprint, 2026. https://doi.org/10.64898/2026.06.19.26356060 (opens in a new tab)