virtual spatial transcriptomics use case
Give molecular data a sense of place.
A bulk RNA profile records what a block of tissue expressed, with no position in it. Sent with the H&E slide from the same block, it informs the predicted spatial expression of that section, and a profile you know well gains a place in the tissue.
For example
In the sample's bulk profile, MS4A1 is one number for the whole block. Predicted on the slide with that profile, it becomes a map, brightest in the lymph node's B-cell follicles.
Worked through for Pharma & Biotech and Biobanks.
The research decision
Should an existing sample-matched bulk RNA profile be included in the spatial prediction?
Combine the slide with a sample-matched bulk RNA profile. The genes, the file and the downstream analysis stay the same.
Where the work gets stuck
Many cohorts already hold a profile
Some cohorts already hold bulk RNA collected for a different research question. When a cohort already pairs bulk RNA with the same tissue block, that profile can provide sample-matched molecular context without another visit to the tissue. The opportunity depends on what the cohort actually holds. Aurora does not assume every laboratory or sample has one.
What it costs you comes to one table: bulk RNA for the same block, in one layout, as the documentation describes. The package that prepares a submission prepares the profile too, and it resolves gene names through the same table every result gets keyed by. Nothing else about the slide changes.
Coordinates are what a bulk profile lacks
One block, two measurements, and for years no way to join them. A bulk profile averages a whole block into one column of numbers, and it records what the tissue expressed. However, it records nothing about where, so on its own it cannot separate the deposit from the node around it. The slide holds the other half and says nothing at all about expression.
Add the sample's bulk RNA profile to the prediction for its slide, and keep the same genes and file format. Without a bulk profile, Aurora predicts from the slide alone.
Because one sample takes one profile, the profile has to belong to the block the slide came from, and the pairing stays yours to get right.
Sample-matched bulk adds context the slide cannot supply
A cohort holding bulk RNA and no spatial data has already paid for half of this. A spatial assay on one sample costs a section and a place in a queue. A bulk profile of the same block costs neither. So a cohort carrying bulk RNA and no spatial data can hand DeepSpot-M a profile it has paid for once, and pay nothing further for it.
What becomes possible
Add the sample's bulk RNA profile to the prediction for its slide, and keep the same genes and file format.
Reuse a profile already available
When the cohort already holds sample-matched bulk RNA, supplying that profile cuts no new tissue and preserves the existing spatial workflow.
Profile-informed maps on the same grid
The figure toggles with and without this sample's bulk profile, same section, same genes. You judge whether the shift helps your question.
The same output file
Neither the panel nor the coordinates change, so nothing downstream needs rewriting to read a result predicted with a profile.
Proof on real tissue
The real-slide example shows the same section and genes predicted with and without its sample-matched bulk profile.
One slide, predicted both ways
Both maps below came from one pass over the same tiles, by DeepSpot-M, on the same day. One input differs between them. Pick a gene, then move between the two. The output structure stays the same.
Every photograph below comes from one open-access diagnostic slide: a melanoma that had spread to a lymph node.
higherYour input and output
- Input
- An H&E slide, and a bulk RNA table for the same block.
- Output
- Virtual spatial transcriptomics at every retained tile, informed by the sample-matched profile.
Take the next step
Start with an example slide, or talk to us about your cohort.