Extending AdsPro to New Surfaces¶

A Major Practical Advantage¶

One of the most important design decisions in AdsPro is how protein-surface interactions are parameterized. Rather than deriving interaction parameters from expensive quantum chemistry or all-atom MD, AdsPro uses experimentally measured free energies of adsorption per residue type — the Morse well-depth ε values. These come from simple chromatography experiments that any biochemistry lab can perform in a day.

This creates an extraordinarily favorable trade-off:

What you need	How you get it	Time
ε per residue type (20 values)	Amino acid retention chromatography	~1 day in the lab
Zeta potential ζ	DLS/ELS measurement	~1 hour
Ionic strength, pH	Your buffer — you already know it	—
The entire simulation pipeline	Already in AdsPro	—

Compare this to all-atom MD: force-field development for the surface, weeks of simulation per protein on HPC clusters, and the entire process repeats for every protein. A full all-atom PMF for one protein on one surface takes weeks to months on a supercomputer cluster.

With AdsPro, you parameterize the surface once from one day of lab work. Every protein in your library can then be simulated in a few hours on a workstation.

Practical impact

For a corona composition study with 20 proteins: AdsPro finishes in ~2 days (including the chromatography experiment). The equivalent all-atom MD study would take years.

The Chromatography Parameterization Protocol¶

The experimental concept is simple: use the same surface chemistry as your NP, packed into a chromatography column, to measure how strongly each amino acid binds.

Step-by-Step Protocol¶

1. Select the right column

Target NP surface	Column type	Main binders expected
Silica (SiO₂, negatively charged)	Q3 silica HPLC	ARG, LYS (electrostatic)
Hydrophobic (polystyrene, carbon nanotube)	Phenyl-Sepharose or butyl-Sepharose (HIC)	TRP, PHE, ILE, LEU, VAL
Gold	Thiol-functionalized beads	CYS (very strong), aromatic residues
Amine-functionalized (positively charged)	Amine-coated column	ASP, GLU (electrostatic)
Carboxyl-functionalized (negatively charged)	Carboxyl-coated column	ARG, LYS (electrostatic)
PEGylated	PEG-coated column	None expected (repulsive)

2. Prepare amino acid solutions

Use nanomolar to low micromolar concentrations. This is the Henry's law regime — the regime where K_H is concentration-independent and equal to the single-molecule partition coefficient. At higher concentrations, surface sites saturate, K_H drops, and the measured ΔG is no longer the single-molecule adsorption energy.

Buffer: match the pH and ionic strength of your target NP experiment (e.g., 50 mM MOPS, pH 7.4, 150 mM NaCl for physiological conditions).

3. Inject and elute

Inject each of the 20 amino acids individually. Measure: - Elution volume V_elution - Column dead volume V_dead (determined with a non-retaining tracer, e.g., deuterium oxide)

4. Compute K_H

\[ K_H = \frac{V_\text{elution} - V_\text{dead}}{V_\text{dead}} \]

(Note: V_net = V_elution − V_dead is the net retention volume.)

5. Convert to ΔG_ads

\[ \varepsilon = \Delta G^\circ_\text{ads} = -RT \ln K_H \]

K_H > 1 → ΔG < 0 → residue binds the surface (ε < 0)
K_H = 1 → ΔG = 0 → no interaction (ε = 0)
K_H < 1 → ΔG > 0 → residue is excluded from the surface (ε > 0)

What Chromatography Gives You (and What It Doesn't)¶

Chromatography measures the depth of the Morse potential well (ε) but not its shape (the α stiffness parameter, currently set to 8 nm⁻¹ universally).

The α parameter controls how sharply the interaction energy decays with distance from the surface. It can be measured from atomic force microscopy (AFM) single-molecule force spectroscopy on the same surface material. However, for most practical purposes, the ε values dominate the physical behavior. A ±20% uncertainty in α changes ΔG_ads by much less than a ±20% uncertainty in ε.

Adding a New Surface to AdsPro¶

Once you have your ε values, adding the surface takes approximately 10 minutes.

Step 1 — Create the surface energy file¶

Create data/aa_parameters/aa_surface_energies_<material>.json:

{
  "_description": "Per-residue adsorption free energies for polystyrene at pH 7.4, 150 mM NaCl",
  "_source_primary": "Your Name (Year). Journal, volume:pages",
  "_experimental_method": "HIC chromatography on phenyl-Sepharose",
  "_conditions": "pH 7.4, 150 mM NaCl, 25 mM phosphate",
  "_morse_alpha": 8.0,
  "_morse_r_min": 0.0,
  "ALA": { "epsilon_kJ_mol":  0.0 },
  "ARG": { "epsilon_kJ_mol":  0.0 },
  "ASN": { "epsilon_kJ_mol":  0.0 },
  "ASP": { "epsilon_kJ_mol":  0.0 },
  "CYS": { "epsilon_kJ_mol": -3.0 },
  "GLN": { "epsilon_kJ_mol":  0.0 },
  "GLU": { "epsilon_kJ_mol":  0.0 },
  "GLY": { "epsilon_kJ_mol":  0.0 },
  "HIS": { "epsilon_kJ_mol": -2.0 },
  "ILE": { "epsilon_kJ_mol": -10.0 },
  "LEU": { "epsilon_kJ_mol": -9.0 },
  "LYS": { "epsilon_kJ_mol":  0.0 },
  "MET": { "epsilon_kJ_mol": -5.0 },
  "PHE": { "epsilon_kJ_mol": -14.0 },
  "PRO": { "epsilon_kJ_mol": -4.0 },
  "SER": { "epsilon_kJ_mol":  0.0 },
  "THR": { "epsilon_kJ_mol":  0.0 },
  "TRP": { "epsilon_kJ_mol": -18.0 },
  "TYR": { "epsilon_kJ_mol": -8.0 },
  "VAL": { "epsilon_kJ_mol": -7.0 }
}

The example above is a hypothetical polystyrene parameterization. Your actual values come from the chromatography experiment.

Step 2 — Register the material in the config manager¶

Open adspro/config_manager.py and add "polystyrene" to the list of known materials:

KNOWN_MATERIALS = {"silica", "polystyrene", "gold", "amine", "carboxyl"}

Step 3 — Register the file path in the energy calculator¶

Open adspro/energy_calculator.py and add the mapping:

SURFACE_ENERGY_FILES = {
    "silica":      "data/aa_parameters/aa_surface_energies.json",
    "polystyrene": "data/aa_parameters/aa_surface_energies_polystyrene.json",
}

Step 4 — Use it in config.yaml¶

nanoparticle:
  material:        polystyrene
  zeta_potential:  -0.005      # V — measure this with DLS for your specific particles
  ionic_strength_mM: 150.0

The entire simulation pipeline — coarse-graining, orientation search, Langevin MD, PMF, WHAM, visualization — runs without any further code changes.

Physical Considerations for Different Surfaces¶

Hydrophobic Surfaces (polystyrene, carbon nanotube)¶

The dominant interaction is the hydrophobic effect — non-polar residues lose unfavorable contacts with water when they adsorb onto a hydrophobic surface. Hydrophobic interaction chromatography (HIC) on phenyl-Sepharose or butyl-Sepharose columns directly measures this.

Expected hierarchy: TRP > PHE > ILE ≈ LEU > VAL > MET (roughly proportional to non-polar surface area).

ARG and LYS would have ε ≈ 0 on polystyrene (they have charged side chains that prefer water). This is the opposite of silica.

Positively Charged Surfaces (amine-functionalized NPs)¶

ASP and GLU (negatively charged at pH 7.4) become the dominant binders through electrostatic attraction. ARG and LYS switch from attractive to repulsive.

Mixed Surfaces (lipid bilayers)¶

Real lipid nanoparticles have a mixed surface — zwitterionic phospholipids (PC head groups, which are neutral), negatively charged lipids (PS, PG), and sometimes PEG chains. Parameterization is more complex but still possible by measuring on vesicle-coated columns.

Validating a New Surface Parameterization¶

After adding a new surface, validate against at least one experimental ΔG_ads measurement before using AdsPro for predictions:

Pick a well-studied protein-NP pair from the literature (e.g., BSA on polystyrene — extensively studied by QCM-D and SPR)
Run AdsPro with your new surface parameters
Compare ΔG_ads to the literature value
If the agreement is within ~5 kJ/mol, proceed with confidence
If discrepant, check: buffer conditions match? pH matches? Protein concentration in the experiment (must be in dilute regime for Henry's law to hold)?