Bradley · Tm
Melting point · K
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
Melting point from a size × polarity-spread FAMILY of exponents (not a single unique pair). The headline self-correction: the published r=0.826 was inflated by a hydrogen-bond classification bug; the honest value is r=0.655.
The physics
Lattice cohesion scales with molecular size and electrostatic complementarity. The data identifies the family (many exponent pairs within ±0.01 r) but not one specific pair — so it ships as a family predictor, not a mechanistic law.
Learning curve is flat (adding 1831 more anchors would buy ≈+0.003 r, below noise) → data-saturated, feature-limited. Ship as a predictor; drop any mechanistic story about the exact exponents.
🚨 CORRECTED 2026-06-06. The v1 published r=0.826 was partially inflated: a 1.2σ H-bond donor threshold miscounted aromatic π-clouds as false donors (~45% false-positive rate on aromatics), and the Bradley cohort is aromatic-rich, so the spurious donor signal leaked into the fit. After the v2 σ-degree atom-role fix, the honest r is 0.655. The other 7 production kernels do NOT select donor/acceptor and are unaffected.
How the methods compare
Melting point is genuinely hard — it needs the crystal lattice, which no cheap method captures. ML wins on MAE (~26–31 K), but by memorising a large training set. This kernel is honest about the ceiling: a size × polarity family at ~38 K MAE, with no training and full interpretability.
Reported error
Each method's own reported error in K — lower is better. Different datasets and splits, so this is an orientation, not a controlled benchmark.
Not charted (no single reported figure): Crystal structure prediction.
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 |
|---|---|---|---|---|---|---|
| Bradley · Tm · this kernel | – | – | – | – | 1 | |
| Group contribution (Joback; Marrero–Gani) | – | – | – | – | – | 0 |
| ML (ASNN / DNN / GNN) | – | – | – | – | 1 | |
| Crystal structure prediction | – | – | – | – | 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 |
|---|---|---|---|---|
| Bradley · Tmthis kernel | closed-form | Pearson r 0.655 (nested-CV) · 0.622 scaffold-blind · MAE 38 K | one DFT SCF σ-profile · no training set · no MD | MF-FQSL (this lab) |
| Group contribution (Joback; Marrero–Gani) | group-contribution | MAE ≳ 30–40 K on diverse sets; fails for rigid/H-bonded solids | 2D functional-group counts | Joback & Reid 1987; Marrero & Gani 2001 |
| ML (ASNN / DNN / GNN) | ML / GNN | MAE ≈ 26–31 K on Bradley / Tetko sets (SOTA, near experimental noise) | 2D descriptors or graph + a large labelled training set | Tetko et al. 2014; Coley et al. (MAE 26.2 K) |
| Crystal structure prediction | physics | No reliable general route — melting needs the crystal lattice | predicted crystal structure + lattice-energy calculation | e.g. Price; CCDC blind tests |
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
- 2026-05-29v1 published (SUPERSEDED)
r=0.826 — later found inflated by the v1 H-bond π-cloud false-positive bug.
- 2026-06-06v2 CORRECTED
σ-degree atom-role classification. Pass A retrobench r=0.610; Pass B (free gate) r=0.655. Form shifted to g_polar+donor. This is the honest number.