2D · positional
Position decides fate
One genome reads two morphogen beads into neuron, epithelial, and stem fates by position. Move a bead, move the boundary.
Wolpert 1969emergent
Open in player Open-science cell simulation
Autonomous cells that differentiate from a single genome, read a morphogen gradient, store a bioelectric memory, commit apoptosis on cue, and metabolize or starve.
An agent-based simulation of developmental biology: autonomous cells whose fate emerges from a Boolean gene-regulatory network, with morphogen signaling and instructive bioelectricity - kinetics-free and deterministic. It reproduces known results within tolerance, re-checked on every change. Free to use, and open to collaboration.
silicospace.com/player
Run spec
genome1
morphogens2
fates3
Within toleranceWolpert 1969 · auto-checked
One genome. Two morphogens. Three emergent cell types. Nothing scripted.
Bit-for-bitdeterministic
2D & 3Dvolumetric runs
μm → weeksreal scales
No signupfree for any lab
Reproduces published results
1969Wolpert1963Steinberg1995Brunner et al.2017Durant, LevinGRNDavidson
The instrument
An open-science instrument, not a black box.
SilicoSpace is built so anyone can check it: the mechanism is legible, the runs are reproducible, and the claims are validated against published biology.
Deterministic and replayable
Same seed, same result, bit for bit. Every run is a protocol you can hand to a collaborator and re-run exactly - an in-silico experiment with a fixed, citable method.
Open methodology
Information flow you can read: genome to gene-regulatory dynamics to fate, signal fields, and voltage. Rate-based, not black-box chemical kinetics, so you can see why each cell decided what it did.
Validated reproductions
The engine reproduces published results as quantitative checks - a number within tolerance, re-run on every change - so a result can never silently break.
Free for any lab
No license, no signup to explore. Bring a hypothesis and get a run you can scrub, replay, recolour, and check at the bench.
Validation
Reproducing the known, on the path to predicting the new.
Each reproduction is a classical or published result the engine must match on a quantitative metric - a number within tolerance, not "looks right" - re-checked on every change so it cannot silently break. With partner data we test it against real systems and use it to explore hypotheses you can check at the bench.
- 01
Morphogen gradient (French flag)
One genome reads a single morphogen gradient into three spatial fate stripes; the emergent stripe boundaries land at the radii the genome's thresholds predict from the closed-form gradient.
Wolpert 1969, positional information
Full page → - 02
Sea-urchin blastula
One Wnt8 vegetal gradient partitions about 1000 cells into ectoderm (~67%), endomesoderm (~28%) and skeletogenic (~4%) territories - the published fate proportions, emergent from the genome.
Davidson endomesoderm GRN; Gilbert, Developmental Biology
Full page → - 03
Differential-adhesion sorting
Two lineages expressing one cadherin at two levels sort from a random mix; the more cohesive lineage is engulfed into an interior core wrapped by a low-cohesion shell.
Steinberg 1963, differential-adhesion hypothesis
Full page → - 04
Density-dependent apoptosis
A dense cluster self-limits: programmed death turns on at the critical size where the cells' superposed death-ligand dose first crosses threshold, while a sparse cluster survives.
Fas-FasL fratricide; Brunner et al. 1995
Full page → - 05
Planaria bioelectric memory
A transient gap-junction blockade yields about 25% two-headed worms whose ectopic voltage pattern persists through amputation - a stored bioelectric memory, not a transient drug effect.
Durant, Levin et al. 2017
Full page → - 06
Tissue from one cell
One founder cell grows into a solid 3D ball whose descendants read a maternal gradient into three concentric shells of fate while contact inhibition self-limits growth - nothing scripted but the genome.
Capstone integration (Wolpert positional info + Kauffman attractors)
Full page →
Every reproduction below runs as an automated quantitative check in the engine's test suite - it fails the build if a result drifts out of tolerance.
In the player
Drive it like an instrument.
The viewer picks the right view from the recording. Orbit a 3D organism, slice it like an MRI, and recolour by the biology - all real engine output.
real engine output
The whole organism, in your hand.
- Drag to rotate the 3D cell cloud; scroll to zoom.
- Cells render as rounded bodies with nuclei, coloured by fate.
- Looks like a morula - because it grew like one.
Explore real experiments
Recordings you can open right now.
Each card is a real recording from the engine - scrub it, replay it, or open it in the full player. Every one credits the work it reproduces.
More to explore
Six more recordings in the player.
Every link opens a real recording in the player - scrub, replay, recolour. The 3D runs open the same player in its orbit + cross-section view.
Tissue from one cell (3D)
3DThe capstone. ONE founder cell divides into a solid ball, and its descendants read a pre-loaded maternal gradient into three concentric spherical shells of fate - a neuron core, an epithelial shell, a stem rim - while contact inhibition self-limits growth. Nothing scripted but the genome and the initial bead. Orbit it, then slice it.
Open in playerSea-urchin blastula (3D)
3DA hollow ball of ~1000 cells. One genome reads a vegetal Wnt8 gradient into ectoderm, endomesoderm and skeletogenic territories - the published fate proportions, emergent. Orbit it, then slice it.
Open in playerFate domains in a sphere (3D)
3DThe two-bead differentiation genome grown as a solid ball instead of a flat sheet: neuron and epithelial domains emerge in 3D around the morphogen sources. Drag to orbit, slide the cutting plane.
Open in playerFrench-flag stripes
A single morphogen gradient read through tiered gene-network thresholds into three clean spatial fate stripes - Wolpert's French flag, grown rather than painted.
Open in playerMany signals at once
A growth-hormone bath, a cell-secreted Shh morphogen and a pulsed stress signal share one dish. Switch the field overlay in the player to see each one.
Open in playerPulsed signal trails
Scattered cells fire transient puffs that spread and fade; two environmental injections send trails across the dish. The overlay shows the signal's past, not just its steady state.
Open in playerHonest scope
What this is not (yet).
Open science means saying what the instrument does not do. Today it reproduces the known; predicting the new is the goal we are working toward with partners.
Read the full capabilities and honest-scope notes- by designNo chemical kinetics. Signaling is rate- and level-based by design, not reaction-rate equations with binding constants.
- frontierThe genome is still abstract gene tokens, not loaded ACGT. The full DNA-to-RNA-to-protein flow is the open frontier.
- abstractionBioelectricity captures the pattern and its memory but abstracts the electrophysiology - no ion-flux or channel kinetics.
- goalPredictive-with-partners is the aim, not a current claim. Treat a run as a hypothesis to check, not an answer.
Collaborate
Bring your system. Build the model together.
SilicoSpace stands on its own as an open instrument - and it sharpens with every real system it models. If a result it produces could be checked at your bench, or a custom model would help your work, bring us the system and we will build and validate it together. Collaboration of any kind is welcome.
For labs
Explore a hypothesis and watch the fate decisions before you commit bench time. Free, no signup.
For PhDs and PIs
Define the validation that matters for your system, co-author the reproduction, and steer what it models next.
For teams and companies
Need a custom model for your system - developmental, regenerative, or bioelectric? We build and validate simulations as a collaboration.
See it move.
Pick a recorded experiment and scrub it live in your browser - no install, no signup.