An engine that invents brain-cancer nanocarriers — and refuses to harm healthy cells.
Most drug-delivery design chases one number: how hard it hits the tumour. This simulator rehearses a different idea — a generative engine that scores every design it dreams up on a therapeutic window: tumour kill weighed against harm to the healthy brain. Watch it learn to invent safer carriers.
In the brain, a toxic miss is permanent
Glioblastoma is among the most lethal cancers. The blood-brain barrier blocks most drugs from reaching it, so carriers like PLGA, Liposomes, Dendrimers, and Gold nanoparticles ferry them across. But the failure that matters most isn't weak tumour kill — it's a carrier that damages healthy neurons, because neuronal loss rarely heals. That is why safety, not raw potency, sits at the centre of every design here.
Configure your nanocarrier or run auto-design
Drag the sliders below to manually engineer the nanocarrier, or use the Generative AI panel to let the engine optimize the properties automatically. The Langevin simulation directly underneath displays the physical behavior of your exact formulation in real-time.
Langevin Microdynamics Sandbox
For a computationally feasible representation of atomic molecular dynamics, we use Langevin/Brownian dynamics. This simulates physical particle advection (blood flow), random thermal diffusion, electrostatic attraction, PEG opsonization, and ligand-receptor binding. Load your highest-scoring TWI champion from the optimizer and watch it behave in real-time.
Rigidity: 45%. PEG coating: 35%.
Hydrophobicity: 55%. Linker: N/A.
Affinity: N/A
- Size: Smaller particles diffuse more chaotically (higher random thermal noise).
- PEGylation: Low PEG leads to rapid immune capture by macrophages (purple blobs).
- Rigidity: High stiffness rebounds off barrier walls, causing shear-damage (flashes red).
- Targeting: Targeted ligands bind to receptor points on the cell surface to cross BBB.
The safety index, defined
Nothing here is hidden behind a black box. The therapeutic-window index is a transparent, dose-resolved quantity — exactly the specification a doctor or methodologist should be able to interrogate.
- K — fraction of tumour cells killed (1 = full kill)
- H — fraction of healthy cells harmed (0 = none)
- σ — the model's own uncertainty at that dose
- λ ≥ 1 — harm-aversion weight (default 2)
- γ ≥ 0 — uncertainty penalty (default 1)
Only positivemargins count — a dose where harm beats kill adds zero, never a bonus elsewhere. So a high score means a genuinely widesafe operating window, not one lucky concentration. A "potent but toxic" carrier — the field's usual winner on raw potency — scores near zero here. That reframe is the whole point.
This is a rehearsal — not a discovery
A claim should be only as strong as the evidence under it. Everything in this simulator lives on the second rung of a three-rung ladder. It shows the method working; it does not show that any carrier is actually safe.
What this can say:"the engine works — it invents novel carriers and optimises a safety objective." What it cannot say:"this carrier is safe" or "we found the causes of nanocarrier safety." A synthetic result reflects the simulator's assumptions, not biology. It is not a substitute for laboratory or clinical judgment.