MNSol · Octanol
Mixed solvation ΔG_solv · kcal/mol
In-sample → out-of-distribution
How the correlation holds when moving to unseen scaffolds.
Vertical marker = the +30% production gate.
Blind calibration slope
Slope of measured vs predicted on the blind set. 1.0 = no magnitude compression.
What it is
Octanol is amphiphilic, so solvation needs both a cavity (size) term and a polar (electrostatic) term. A 30-seed bootstrap shows the FORM is stable (selected 89% of resamples); the magnitude is noisy from a small blind cohort.
The physics
Mixed polar/non-polar solvent → both hydrophobic cavity opening and electrostatic screening matter. The two-term form is physically real.
Published nested-CV r=0.867 is the full-data upper bound; the honest blind estimate is ≈0.66. The 1-term fallback is the conservative floor.
Bootstrap (30 seeds): cavity+g_polar chosen 89% of folds, H-bond features never selected. But n_blind median = 8 → r_blind IQR is wide (std 0.236). Form real, magnitude uncertain — widen the CI, don't over-trust the point estimate.
How the methods compare
The physics gold standard — alchemical FEP — reaches ~1 kcal/mol but needs a full molecular-dynamics simulation per molecule. SMD and COSMO-RS need a DFT job. This kernel lands in the same order of magnitude from a single closed-form solve on one σ-profile: it trades a little accuracy for orders of magnitude less compute, and stays fully interpretable.
Reported error
Each method's own reported error in kcal/mol — lower is better. Different datasets and splits, so this is an orientation, not a controlled benchmark.
Lower error isn’t the whole story — read it against input cost and requirements below.
Input cost
Per-molecule compute/data burden to predict a new molecule — shorter is cheaper.
What each method needs
External dependencies each method carries. An amber dot means the method requires it — fewer dots means fewer things to procure or that can go wrong.
| Method | 3D geometry | MD / sampling | training corpus | a measured value | proprietary params | deps |
|---|---|---|---|---|---|---|
| MNSol · Octanol · this kernel | – | – | – | – | 1 | |
| Alchemical FEP (explicit-solvent MD) | – | – | – | 2 | ||
| SMD (implicit-solvent DFT) | – | – | – | – | 1 | |
| COSMO-RS / COSMOtherm | – | – | – | 2 | ||
| D-MPNN (Chemprop) | – | – | – | – | 1 |
This kernel needs only a 3D geometry for its one SCF — no MD, no training corpus, no measured value, no proprietary software.
Competing methods
How this property is predicted elsewhere — with the input each method needs (a key differentiator) and the literature reference. Numbers are each method’s own reported figure on its own benchmark, so they are indicative, not a head-to-head on an identical split.
| Method | Class | Reported performance | Input needed | Reference |
|---|---|---|---|---|
| MNSol · Octanolthis kernel | closed-form | Pearson r 0.867 (nested-CV) · 0.663 scaffold-blind · MAE 1.09 kcal/mol | one DFT SCF σ-profile · no training set · no MD | MF-FQSL (this lab) |
| Alchemical FEP (explicit-solvent MD) | physics | MAE 1.07, RMSE 1.43 kcal/mol (GAFF/TIP3P, FreeSolv n=621) | 3D geometry + force field + molecular-dynamics sampling | Mobley & Guthrie, FreeSolv, J. Comput. Aided Mol. Des. 2014 |
| SMD (implicit-solvent DFT) | physics | MUE 0.6–1.0 kcal/mol for neutrals (6-31G*); ~2.5 kcal/mol over full MNSol | DFT SCF + continuum, per solute | Marenich, Cramer & Truhlar, J. Phys. Chem. B 2009 |
| COSMO-RS / COSMOtherm | physics | ≈0.4–0.6 kcal/mol on well-parametrised sets (best-in-class physics) | DFT σ-profile + proprietary parametrisation | Klamt, J. Phys. Chem. 1995 |
| D-MPNN (Chemprop) | ML / GNN | RMSE ≈ 1.2 kcal/mol on FreeSolv (MoleculeNet split) | 2D molecular graph + a labelled training set | Yang et al., J. Chem. Inf. Model. 2019 |
Metrics are as published by each method on its own dataset (different splits, different cohorts) — treat them as an orientation of the landscape, not a controlled benchmark. The differentiator for this kernel is the input column: a single closed-form solve from one σ-profile, with no training corpus, no MD, and no measured melting point.
Descriptors used
Version history
- 2026-05-29v0.91.1 ship
Nested-CV r=0.867 (cavity+g_polar).
- 2026-06-06Bootstrap clarification
30-seed bootstrap: form stable (89%), magnitude noisy (n_blind≈8, std 0.236). NOT classic overfit — sampling variance. Verdict: widen CI, do not retire.