See new possibilities for the medicines of tomorrow.

Applications

A broader field of opportunity for discovery

1. Explore

   Revisit a completed study with a new question

   The slides a finished study left behind can take up a question it was never designed to answer, including once its tissue is exhausted or reserved for other work.

   • Compare historical cohorts on the same predicted genes
   • Explore a question the original study never asked
   • Take candidate findings to validation in current cohorts

   Built onOut-of-the-box prediction · Comparative spatial analysis

   DeepSpot-M produced a virtual spatial atlas of 28,664 slides across 32 cancer types, published on Hugging Face.^{[1, 2]}

2. Explore

   Follow unexpected connections across the whole cohort

   Samples from the same indication can carry very different tissue, while molecular characterisation usually reaches a subset of them.

   Predicted expression on every slide sets the profiled cases beside the rest, and a biomarker hypothesis can be explored where it sits in the tissue.

   • Compare predicted expression across every sample
   • Propose subgroups of patient samples for research and hypothesis generation
   • Find where a candidate gene's predicted expression runs highest
   • Add a sample-matched bulk profile to the prediction where one exists

   Built onOut-of-the-box prediction · Comparative spatial analysis · Sample-matched bulk RNA · The analysis report and the explorer

   The web upload does not take a bulk profile. It goes in through the Python package and the API, on the Commercial route.See Commercial

   DeepSpot-M predicts 19,338 protein-coding genes at every retained tile of a slide, from the H&E image alone.^[1]

3. Extend

   Take a study beyond the panel it was designed around

   With our team, a completed spatial study gains predictions for the protein-coding genes its panel left out, adapted to the data it measured.

   The predicted genes stay labelled beside the measured ones, so a question about the tissue around a target can go past the edge of the panel.

   • Examine the genes around a target, off the panel
   • Restore genes lost to dropout or quality control, as labelled predictions
   • Choose which off-panel genes move into validation

   Built onPanel extension · Finetune · Restore genes

   Arranged with our team, on the Advanced R&D route.See Advanced R&D

   The DeepSpot-M preprint demonstrates transfer to held-out cancers, and adaptation to a new cohort from one slide.^[1]

4. Look ahead

   Begin each new study with a wider view

   Predicted expression across the candidate samples shows where the biology a study is about sits, before any tissue is cut for it.

   • Rank candidate samples on the genes the question depends on
   • Aim each capture area at the region the hypothesis turns on
   • Choose a targeted panel after looking at predicted expression

   Built onOut-of-the-box prediction · Comparative spatial analysis

Worked examples

See each direction on real tissue

• 

  A new chapter for your collection.

  Explore the whole archive on the same predicted genes, and keep measured follow-up for the hypotheses that survive.

  You bring

  One H&E slide per sample. Nothing else.

  Open the worked example
• 

  Give molecular data a sense of place.

  Add the sample's bulk RNA profile to the prediction for its slide, and keep the same genes and file format.

  You bring

  An H&E slide, and a bulk RNA table for the same block.

  Open the worked example
• 

  Begin your next experiment with a wider view.

  Rank the candidate samples on predicted expression, and choose where each capture area goes.

  You bring

  The slides you are choosing between.

  Open the worked example