DNA unlocks the gene vocabulary.
DNA-derived identities bring coding and non-coding genes into one shared language for modelling. Each cell’s expression values fill fixed gene positions to form a transcriptome image.
THE CELLNET FILM · 44 SECONDS
Panoramic transcriptome modelling
Preparing the film…
THE BIOLOGICAL POSSIBILITY
Explore the unmeasured.
Reveal biological connections.
The panorama, within reach.
Image patches group many genes into a compact token grid, making panoramic transcriptome modelling computationally practical.
01
02
03
See the rare. Resolve the state.
Distinguish rare cell populations and condition-associated states. Connect cellular identity to the biology of tissues.
Explore beyond measured genes.
Expand the gene space available for biological exploration. Predict expression beyond an assay’s measured genes to investigate cell states and spatial expression patterns.
Non-coding RNA. Functional connections.
Start with non-coding RNA and retrieve associated gene programmes. Connect underexplored transcripts to developmental and neural biology.
OPEN FOR DISCOVERY
Image Atlas
An open, large-scale image collection. Explore cellular diversity and build on a shared visual foundation.
A MODEL FOR DISCOVERYFoundation Model
Gene identity encoded in DNA. Practical panoramic learning through images. Explore expression and functional programmes in cellular context.