MultiscaleNano

Simulation methodology

MultiscaleNano runs real molecular dynamics (OpenMM) for coarse-grained modules and applies documented physical bridges for continuum transport and release. Every completed run includes per-metric equations, references, and 95% confidence intervals from independent MD replicates where applicable.

Simulation modes

Standard MD

5,000 MD steps · 3 replicates

Production MD

25,000 MD steps · 5 replicates

Pipeline modules

Encapsulation

Martini 3 (OpenMM) or coarse-grained LJ fallback

  • Potential energy from Langevin MD trajectory
  • Encapsulation efficiency from core bead fraction × compactness
  • Drug retention ΔG from mean potential energy per bead

Replicates: 3 (standard) / 5 (production)

Formation

Martini 3 lipid self-assembly (OpenMM)

  • Hydrodynamic radius R_H ≈ 2 × R_g from final configuration
  • Morphology classification (core-shell vs compact-sphere)
  • Polydispersity from replicate spread

Replicates: 3 / 5

Stability

Martini 3 thermal perturbation (+10 K)

  • Stability score from R_g change under heat stress
  • Drug leakage rate from energy fluctuation proxy

Replicates: 3 / 5

Transport

Stokes–Einstein continuum bridge

  • Effective diffusion D = k_B T / (6πηR_H)
  • Penetration depth x ≈ √(2Dεt) in porous tissue

Replicates: Deterministic from formation MD

Release

Slab diffusion with MD-derived D_eff

  • Half-life t_½ = R_H² / (2D_eff)
  • Release profile from exponential cumulative model

Replicates: Deterministic from stability + formation

Protein corona

OpenMM competitive adsorption MD

  • Adsorbed protein count within 1.2 nm of particle surface

Replicates: 3 / 5

Cell interaction

OpenMM NP–membrane approach MD

  • Membrane adhesion energy (kT)
  • Uptake and endosomal escape probabilities from adhesion

Replicates: 3 / 5

Limitations

Lipid modules use the Martini 3 coarse-grained force field (via martini_openmm and insane) when available — the same class of model used on nanoHUB GROMACS workflows. Ionizable lipids (SM-102, ALC-0315, MC3) map to closest Martini lipid templates until custom topologies are added. Corona and membrane modules still use the legacy LJ bead model. Results are computational predictions and should supplement experimental validation.