Methods Overview¶
AdsPro is built on four scientific pillars:
- Coarse-grained protein model — one bead per residue, pH-dependent charges, Gō native contact model
- Physics-based surface model — Fibonacci-sphere NP discretization, Grahame-equation charge calibration, Debye-Hückel electrostatics
- Experimentally parameterized force field — Morse surface potential from chromatography, Gō model for internal energy
- Rigorous free energy calculation — Langevin MD + steered MD + umbrella sampling + WHAM
Together these enable computing ΔG_ads for any protein on any parameterized surface in hours rather than the weeks required by all-atom MD.
Methods Pages¶
| Page | Contents |
|---|---|
| Protein Model | Cα coarse-graining, Henderson-Hasselbalch charge assignment, Gō contact map |
| Surface Model | Fibonacci sphere placement, Grahame equation, Debye length |
| Force Field | Morse potential, Gō model, Debye-Hückel, backbone restraints |
| Simulation Protocol | Langevin dynamics, velocity-Verlet integration, Phase 0 / Phase 1 |
| PMF & WHAM | Steered MD, smart window detection, umbrella sampling, WHAM |
Level of Theory¶
AdsPro is a coarse-grained implicit-solvent model. This means:
- Implicit solvent — water is not simulated explicitly; its effect enters via the dielectric constant ε_r = 78.4 and the Debye screening length λ_D
- One-bead-per-residue resolution — each amino acid is a single sphere at the Cα position; internal backbone geometry is frozen (harmonic springs maintain bond lengths)
- Gō-model internal energy — only native contacts are attractive; the protein naturally wants to stay folded
- Empirical surface potential — Morse well depths come from single amino acid chromatography experiments, not from quantum chemistry
This level of theory is justified because:
- The key observable — adsorption free energy ΔG_ads — is dominated by interactions at the protein-surface interface, which the Morse + Debye-Hückel model captures correctly
- At the length scales of interest (1–10 nm), continuum electrostatics (Debye-Hückel) is a well-validated approximation
- The Gō model correctly captures the competition between surface binding and native fold stability — the biologically critical quantity
- Reduction from ~20,000 atoms (all-atom) to ~100–200 beads (CG) enables umbrella sampling that would be computationally intractable at full atomic resolution
Comparison to All-Atom MD¶
| Aspect | All-atom MD | AdsPro CG-MD |
|---|---|---|
| Resolution | Every atom (~20,000 for Lysozyme) | One bead/residue (~129) |
| Time step | 2 fs | 10 fs |
| Force field | AMBER/CHARMM + TIP3P water | Gō + Morse + Debye-Hückel |
| PMF timescale needed | μs–ms | ns |
| Wall-clock time per protein | Weeks–months (HPC) | Hours (workstation) |
| New surface cost | Weeks of force-field development | 1 day chromatography experiment |
| ΔG_ads accuracy | ~1–2 kJ/mol (well-converged) | ~3–5 kJ/mol (CG approximation) |
The ~3–5 kJ/mol accuracy of AdsPro is sufficient for protein corona composition studies, where the key question is relative affinity ranking rather than absolute energetics.