Platform · Formulation Engine

Screen 10,000 formulations before touching a pipette

Our physics-informed model predicts ionizable lipid pKa, encapsulation efficiency, and particle size for each candidate — from first principles, not curve fitting.

How It Works

Physics-informed, not black-box

The Formulation Engine uses a coarse-grained molecular dynamics model parameterized against published lipid force fields. Each candidate formulation is simulated at physiological pH to predict:

  • Ionizable lipid pKa — endosomal protonation propensity that governs release kinetics
  • Encapsulation efficiency — predicted mRNA retention within the lipid core
  • Particle size distribution — hydrodynamic diameter and polydispersity index estimate
  • Membrane fusion propensity — endosomal membrane perturbation likelihood at pH 5.5
Formulation Parameter Grid — Sample Output
Formulation IL mol% PEG mol% pKa pred. EE pred. Z-avg nm Fus. score
FE-0042501.56.4288%1030.87
FE-0117502.06.3185%1140.83
FE-0058401.56.7581%980.74
FE-0203552.56.1879%1270.68
FE-0301451.07.1074%890.61

Showing 5 of 10,000 screened formulations. Amber rows: pKa 6.2–6.5, fusion score >0.80 — prioritized for bench validation.

Design Space

What we screen

Each campaign explores a combinatorial design space across four primary formulation axes.

Ionizable Lipids

Library of 1,200+ ionizable lipid structures, parameterized by pKa, headgroup geometry, and acyl chain length.

Molar Ratios

IL:Helper:Chol:PEG molar ratio combinations across a grid centered on known efficacious formulations.

PEG-Lipid Variants

PEG2000-DMG, PEG2000-DSG, PEG1000 variants at densities from 0.5–3.5 mol% screened for corona formation and stealth tradeoffs.

Helper Lipids

DSPC, DOPE, DPPC helper lipid variants screened for impact on membrane fluidity and endosomal release kinetics.

Ready to run your formulation campaign?

Tell us your target tissue, payload size, and route of administration — we handle the rest.