Three connected programmes, all built on one idea: a working model of a living cell that you can run, test, and learn from. Everything here is research — published openly, and free for anyone to use.
THE CORE
Physics-inspired classical and quantum digital twins of a living cell — the core of everything we build, wrapped in a multi-agent system that puts them to work.
A cell is not a list of measurements. It is a system that moves: genes switch on, proteins are made and cleared, energy is spent and replenished. Quantum Virtual Omics (QVO) is our effort to capture that motion in software — a digital twin of a cell that behaves the way a real one does, built from measurements of what is actually inside it.
What makes it different from a statistical fit is that the basic rules of biology are built into the model rather than learned by accident. Matter is not created or destroyed; energy has to come from somewhere. Because those constraints hold by construction, the twin stays physically sensible even when asked about conditions it has not seen.
A model on its own does not answer a research question. Around the QVO we are building a layer of specialist AI agents — software that finds and prepares the data, fits and checks the model, runs the what-if studies, and turns results into something a researcher can act on. The QVO supplies a cell state you can interrogate; the agents do the work of getting data to it and reading its answers back out. Together they make an end-to-end stack rather than a model you have to drive by hand.
The cell model is in active development with manuscripts in preparation; the agent layer is specified and being built.
Classical and quantum, side by side
We build both. The physics-inspired structure keeps each model honest — matter and energy have to balance — and the quantum version uses quantum computing as the substrate for carrying many coupled variables at once. Running the two together lets us see exactly where each approach earns its keep, which is the interesting question and the one we are working on now.
The data we use
Public, published datasets rather than data we hold privately: mouse-liver circadian time-series covering three molecular layers — transcripts, proteins, and metabolites — drawn from GEO, a published SILAC proteomics study, and the Metabolomics Workbench. Anyone can obtain the same inputs and check what we report.
Where this stands
Both a classical and a quantum version of the cell model are built and running on real, publicly available multi-omic time-series data, developed as a matched pair so the two can be compared directly. Manuscripts on both are in preparation. The quantum side runs on simulators today, with hardware the next step as it becomes available. The agent layer is specified and being built. New results are arriving steadily.
What this involves
BUILT ON THE CORE
The human body has more than 250 kinds of cell. They have more in common than you might expect.
A heart cell and a liver cell look nothing alike and do entirely different jobs. But underneath, both run the same housekeeping: reading genes, making proteins, managing energy, clearing waste. What makes one a heart cell is a small set of extra capabilities layered on that shared foundation.
The atlas builds that idea directly on the QVO. We start from a generic cell and describe each real cell type as the generic one plus a few added specialities — contraction, secretion, electrical signalling, and so on. Because every cell type is described in the same vocabulary, they can be compared honestly, and the in-between states a cell passes through as it develops become describable too.
Where this stands
Underway. The reference cell and the block library are being assembled now, with broader coverage of cell types following as the library grows.
What this involves
BUILT ON THE CORE
If a healthy cell is a working model, disease is that model running differently.
Most disease classification works by pattern matching: this profile resembles that label. Because the QVO is a model of how a healthy cell behaves, we can ask a sharper question. We are exploring a different question — what actually changed? Given a model of how a healthy cell behaves, a disease shows up as a specific alteration to it. Something became too strong or too weak; a connection was lost; waste is cleared too slowly.
The value of describing it that way is comparison. Cancer, metabolic disease, immune conditions, and neurodegeneration all get described in the same terms, so they can be studied side by side instead of in separate silos.
Where this stands
Early stage, and open from the start. We publish the approach as we develop it so others can build on it, test it, and tell us where it needs work.
What this involves
Research scope
This is open scientific research. The Foundation does not provide medical care, diagnosis, or treatment, and nothing here is medical advice. Our work uses public, synthetic, and consented research data. Any clinical use would need separate validation, ethical review, and regulatory approval — work we do not undertake.
