🧪 Live Experiment Console

Inject a perturbation → the differentiable cell responds → measure. Every number is our model's real output.

③ Measure — live readout

ATP productiondifferentiable whole-cell model

••• relative

ATP yield

1.94

Doubling

117m

ori : ter

1.24

Validated vs public syn3A: ori:ter 1.24 vs 1.21, doubling 117 vs 105 min — grade A, no fitting.

① Target∂ backward · inverse design

ENO🔒

PGK🔒

NOX🔒

ACKr🔒

The model computes ∂ATP / ∂enzyme by autodiff through the whole cell (FD-verified to 10.6 digits). It ranks the highest-leverage target — a forward-only simulator cannot.

② Inject & run▶ forward · simulate

💊 Inhibit ENO

↺ Reset

Baseline steady state · ATP energy (relative). Pick a target and inject.

Enzyme barrier from quantum (VQE/DFT) energetics → Eyring kcat → whole-cell steady state. No real quantum hardware run (θ is a placeholder); the coupling & sensitivity are FD-verified.

🦠 E. coli — Min system

A self-organizing protein oscillation that tells the cell where to divide — the textbook case where space matters.

Oscillation period

41.7 s

MinD sweeps pole-to-pole; its time-average is minimal at midcell → marks the division plane. Inside the measured 40–120 s range — predicted, not fit (Huang-Meir-Wingreen 2003).

Why it's ours∂ differentiable

We differentiate through the pattern (autodiff vs FD, 7–8 digits) → inverse-design a spatial pattern. Stochastic 4D simulators cannot. Physics gates: grade A.

🔴 Honest: deterministic mean-field (real Min is stochastic, ~4000 molecules) · 1D long-axis · published params · the Min subsystem, not a whole E. coli cell.

VIDRAFT · Differentiable Cell Platform

One differentiable engine —
from a minimal cell to human disease to our own drugs

Every figure below is our model's own output — forward-simulated, gradient-verified, graded against public data where it exists, with honest limits stated. Self-correcting agentic research.

🧫 Minimal cell→🦠 Bacteria→🧠 Human disease→💊 Our drugs→🫁 Organ metabolism

🧫

JCVI-syn3A minimal cell

Differentiable whole-cell of the smallest genome (493 genes); it replicates & divides.

ori:ter 1.24 vs 1.21 measuredA

No fitting. Prototype; deterministic, not 4D-stochastic.

🦠

E. coli — Min oscillation

Pole-to-pole protein wave that sets the division site; a differentiable spatial pattern.

period 41.7 s in 40–120 s rangeA

Published model, not re-fit. Mean-field, not stochastic.

🧠

Human disease — protein aggregation

Differentiable neurodegeneration aggregation cascades (Alzheimer-type), with mechanism-based screening.

validated vs published kinetics A

Parameter-free scaling law reproduced (no fit); mechanism-plausibility, not efficacy.

🎯

Drug mechanism screening

Gradients rank which molecular step a therapy must hit — and flag counterproductive ones — across our disease models.

∂(outcome)/∂(mechanism) FD-verified

Capability shown on public targets; effects are uncertainty bands, not efficacy. Specific compounds & targets are private.

🫁

Genome-scale human metabolism

Differentiable human liver metabolism on the real Recon3D reconstruction (10,600 reactions); fatty-liver phenotype emerges from mass balance.

8/8 liver checks · Recon3DA

Constraint-based metabolism, not a whole cell; pinpoints the highest-leverage disease step.

🕰️

Universal clock (reproduction)

Reproduced the equant "universal dynamical clock" method; verified on a Kepler orbit.

31,531× more uniform null on ours

Honest null: no gain on our symmetric oscillators. Core reproduced, not the full paper.

🔒 Private R&D preview · research prototype · mechanism-plausibility models, not efficacy or clinical predictions · method patent-pending.

VIDRAFT · Virtual Cell Platform

Not a simulation you watch.
A cell you can run backward.

JCVI-syn3A — the smallest genome that builds a living cell — rebuilt inside a GPU as a differentiable model. Run it forward to simulate life; run it backward to compute exactly what to change.

493-gene genome · differentiable end-to-end

ori:ter 1.24 vs 1.21 measured — grade A, no fitting

quantum→cell gradients · 10.6-digit verified

🗺️

Mechanistic whole-cell (SOTA)

A precise map. But static — to ask "what if?" you re-run blindly. Not differentiable, no inverse design.

🧭

AI cell models

A compass with no map — data-driven, no real physics. Struggles to beat a linear baseline on perturbations.

🛰️

VIDRAFT — map + GPS

Real mechanism you can differentiate: it tells you which knob to turn and re-routes instantly. Forward AND backward.

Enter the cell ▶

Research prototype · differentiable minimal-cell method patent-pending · every figure is our own validated model output.

The DifferentiableVirtual Cell

One differentiable engine —from a minimal cell to human disease to our own drugs

Not a simulation you watch.A cell you can run backward.

The Differentiable
Virtual Cell

One differentiable engine —
from a minimal cell to human disease to our own drugs

Not a simulation you watch.
A cell you can run backward.