Lipinski Rule of Fives vs Congreve Rule of Threes — Reference Table
P0 reference — verbatim Table 5.1 from 2014-schneider-de-novo-molecular-design-book §5.2.2 (Durrant & Amaro, p. ~127). The drug-likeness/fragment-likeness filters used at virtually every fragment-based drug-discovery (FBDD) pipeline entry, including h01 phase 3b fragment scoring. Cited where:
- Lipinski et al., Adv. Drug Deliv. Rev. 46 (2001) 3–26 [12]
- Congreve, Carr, Murray, Jhoti. Drug Discov. Today 8(19) (2003) 876–877 [16]
Verbatim Table 5.1
| Property | Rule of fives (drugs) | Rule of threes (fragments) |
|---|---|---|
| Molecular mass | ≤500 Da | ≤300 Da |
| Hydrogen-bond acceptors | ≤10 | ≤3 |
| Hydrogen-bond donors | ≤5 | ≤3 |
| Partition coefficient (clogP) | ≤5.0 | ≤3.0 |
| Rotatable bonds | — | ≤3 |
| Polar surface area | — | ≤60.0 Ų |
— Verbatim from Schneider 2014, Table 5.1, Ch. 5 (Durrant & Amaro).
How to use in STRC
- h01 fragment library generation (phase 3b): apply rule-of-three as the upper envelope on heavy-atom variants; do not impose lower bounds — fragments shrink during pocket-fit. The 180–350 Da pocket-fit window in
phase3b.py(per pharmacochaperone) is narrower than rule-of-three’s ≤300 Da bound and reflects in-silico calibration to the 159 ų E1659A subpocket. - Citation: when a docstring claims “Congreve rule-of-three filter,” cite both Congreve 2003 (original rule) and Schneider 2014 §5.2.2 (textbook treatment). The two-citation pattern matches the audit fix already applied in
[[literature/pharmacochaperone]]. - Drug stage (phase 7+): elaborated leads must satisfy rule-of-five. Optimization typically adds 80–150 Da to fragment hits (Schneider 2014 §5.2.4), so a ≤300 Da fragment optimized to ≤500 Da drug has only ~200 Da of headroom — confirms the discipline of starting low.
- Solubility constraint not in this table: Congreve’s original rule mentions H-bond donors/acceptors only; the ≤60 Ų polar surface area and ≤3 rotatable bonds are optional additions (per Schneider 2014 §6.2.2). Aqueous solubility ≥1 mM (Mazanetz library) or ≥2 mM (Vernalis) is a separate filter — see Recipe — Fragment Library Filtering Pipeline.
- Fragment hit-rate impact: of 21.1 M commercial compounds (Evotec EVOsource 2014), only ~94,000 satisfy rule-of-three plus the optional TPSA/rotatable-bond constraints (Schneider 2014 §6.2.5, citing Zuegg & Cooper 2009). The filter is severely restrictive — useful when discipline is needed.
Connections
[part-of]pharmacochaperone[source]2014-schneider-de-novo-molecular-design-book[applies]index[see-also]Recipe — Fragment Library Filtering Pipeline[see-also]Ligand Efficiency Metrics Catalog